jusax23/flutter_stockfish_plugin
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

full project

Browse source
This commit is contained in:
jusax23 2023-10-29 22:47:21 +01:00
parent dbc707f23f
commit 884c258669
Signed by: jusax23
GPG key ID: 499E2AA870C1CD41
84 changed files with 21232 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

46
.gitignore vendored Normal file
View file

@ -0,0 +1,46 @@
# Miscellaneous
*.class
*.log
*.pyc
*.swp
.DS_Store
.atom/
.buildlog/
.history
.svn/
migrate_working_dir/
# IntelliJ related
*.iml
*.ipr
*.iws
.idea/
# The .vscode folder contains launch configuration and tasks you configure in
# VS Code which you may wish to be included in version control, so this line
# is commented out by default.
#.vscode/
# Flutter/Dart/Pub related
# Libraries should not include pubspec.lock, per https://dart.dev/guides/libraries/private-files#pubspeclock.
/pubspec.lock
**/doc/api/
.dart_tool/
build/
#/src/Stockfish
trash
.github
.vscode
# Files from build
**/*.o
**/*.s
src/.depend
# Built binary
src/stockfish*
src/-lstdc++.res
# Neural network for the NNUE evaluation
**/*.nnue

27
.metadata Normal file
View file

@ -0,0 +1,27 @@
# This file tracks properties of this Flutter project.
# Used by Flutter tool to assess capabilities and perform upgrades etc.
#
# This file should be version controlled and should not be manually edited.
version:
revision: "476aa717cd342d11e16439b71f4f4c9209c50712"
channel: "beta"
project_type: plugin
# Tracks metadata for the flutter migrate command
migration:
platforms:
- platform: root
create_revision: 476aa717cd342d11e16439b71f4f4c9209c50712
base_revision: 476aa717cd342d11e16439b71f4f4c9209c50712
# User provided section
# List of Local paths (relative to this file) that should be
# ignored by the migrate tool.
#
# Files that are not part of the templates will be ignored by default.
unmanaged_files:
- 'lib/main.dart'
- 'ios/Runner.xcodeproj/project.pbxproj'

5
CHANGELOG.md Normal file
View file

@ -0,0 +1,5 @@
## 1.0.0
First Release:
* Usable
* Stockfish logger is not usable due to cout/cin stream manipulationen.

674
LICENSE Normal file
View file

@ -0,0 +1,674 @@
GNU GENERAL PUBLIC LICENSE
Version 3, 29 June 2007
Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
Everyone is permitted to copy and distribute verbatim copies
of this license document, but changing it is not allowed.
Preamble
The GNU General Public License is a free, copyleft license for
software and other kinds of works.
The licenses for most software and other practical works are designed
to take away your freedom to share and change the works. By contrast,
the GNU General Public License is intended to guarantee your freedom to
share and change all versions of a program--to make sure it remains free
software for all its users. We, the Free Software Foundation, use the
GNU General Public License for most of our software; it applies also to
any other work released this way by its authors. You can apply it to
your programs, too.
When we speak of free software, we are referring to freedom, not
price. Our General Public Licenses are designed to make sure that you
have the freedom to distribute copies of free software (and charge for
them if you wish), that you receive source code or can get it if you
want it, that you can change the software or use pieces of it in new
free programs, and that you know you can do these things.
To protect your rights, we need to prevent others from denying you
these rights or asking you to surrender the rights. Therefore, you have
certain responsibilities if you distribute copies of the software, or if
you modify it: responsibilities to respect the freedom of others.
For example, if you distribute copies of such a program, whether
gratis or for a fee, you must pass on to the recipients the same
freedoms that you received. You must make sure that they, too, receive
or can get the source code. And you must show them these terms so they
know their rights.
Developers that use the GNU GPL protect your rights with two steps:
(1) assert copyright on the software, and (2) offer you this License
giving you legal permission to copy, distribute and/or modify it.
For the developers' and authors' protection, the GPL clearly explains
that there is no warranty for this free software. For both users' and
authors' sake, the GPL requires that modified versions be marked as
changed, so that their problems will not be attributed erroneously to
authors of previous versions.
Some devices are designed to deny users access to install or run
modified versions of the software inside them, although the manufacturer
can do so. This is fundamentally incompatible with the aim of
protecting users' freedom to change the software. The systematic
pattern of such abuse occurs in the area of products for individuals to
use, which is precisely where it is most unacceptable. Therefore, we
have designed this version of the GPL to prohibit the practice for those
products. If such problems arise substantially in other domains, we
stand ready to extend this provision to those domains in future versions
of the GPL, as needed to protect the freedom of users.
Finally, every program is threatened constantly by software patents.
States should not allow patents to restrict development and use of
software on general-purpose computers, but in those that do, we wish to
avoid the special danger that patents applied to a free program could
make it effectively proprietary. To prevent this, the GPL assures that
patents cannot be used to render the program non-free.
The precise terms and conditions for copying, distribution and
modification follow.
TERMS AND CONDITIONS
0. Definitions.
"This License" refers to version 3 of the GNU General Public License.
"Copyright" also means copyright-like laws that apply to other kinds of
works, such as semiconductor masks.
"The Program" refers to any copyrightable work licensed under this
License. Each licensee is addressed as "you". "Licensees" and
"recipients" may be individuals or organizations.
To "modify" a work means to copy from or adapt all or part of the work
in a fashion requiring copyright permission, other than the making of an
exact copy. The resulting work is called a "modified version" of the
earlier work or a work "based on" the earlier work.
A "covered work" means either the unmodified Program or a work based
on the Program.
To "propagate" a work means to do anything with it that, without
permission, would make you directly or secondarily liable for
infringement under applicable copyright law, except executing it on a
computer or modifying a private copy. Propagation includes copying,
distribution (with or without modification), making available to the
public, and in some countries other activities as well.
To "convey" a work means any kind of propagation that enables other
parties to make or receive copies. Mere interaction with a user through
a computer network, with no transfer of a copy, is not conveying.
An interactive user interface displays "Appropriate Legal Notices"
to the extent that it includes a convenient and prominently visible
feature that (1) displays an appropriate copyright notice, and (2)
tells the user that there is no warranty for the work (except to the
extent that warranties are provided), that licensees may convey the
work under this License, and how to view a copy of this License. If
the interface presents a list of user commands or options, such as a
menu, a prominent item in the list meets this criterion.
1. Source Code.
The "source code" for a work means the preferred form of the work
for making modifications to it. "Object code" means any non-source
form of a work.
A "Standard Interface" means an interface that either is an official
standard defined by a recognized standards body, or, in the case of
interfaces specified for a particular programming language, one that
is widely used among developers working in that language.
The "System Libraries" of an executable work include anything, other
than the work as a whole, that (a) is included in the normal form of
packaging a Major Component, but which is not part of that Major
Component, and (b) serves only to enable use of the work with that
Major Component, or to implement a Standard Interface for which an
implementation is available to the public in source code form. A
"Major Component", in this context, means a major essential component
(kernel, window system, and so on) of the specific operating system
(if any) on which the executable work runs, or a compiler used to
produce the work, or an object code interpreter used to run it.
The "Corresponding Source" for a work in object code form means all
the source code needed to generate, install, and (for an executable
work) run the object code and to modify the work, including scripts to
control those activities. However, it does not include the work's
System Libraries, or general-purpose tools or generally available free
programs which are used unmodified in performing those activities but
which are not part of the work. For example, Corresponding Source
includes interface definition files associated with source files for
the work, and the source code for shared libraries and dynamically
linked subprograms that the work is specifically designed to require,
such as by intimate data communication or control flow between those
subprograms and other parts of the work.
The Corresponding Source need not include anything that users
can regenerate automatically from other parts of the Corresponding
Source.
The Corresponding Source for a work in source code form is that
same work.
2. Basic Permissions.
All rights granted under this License are granted for the term of
copyright on the Program, and are irrevocable provided the stated
conditions are met. This License explicitly affirms your unlimited
permission to run the unmodified Program. The output from running a
covered work is covered by this License only if the output, given its
content, constitutes a covered work. This License acknowledges your
rights of fair use or other equivalent, as provided by copyright law.
You may make, run and propagate covered works that you do not
convey, without conditions so long as your license otherwise remains
in force. You may convey covered works to others for the sole purpose
of having them make modifications exclusively for you, or provide you
with facilities for running those works, provided that you comply with
the terms of this License in conveying all material for which you do
not control copyright. Those thus making or running the covered works
for you must do so exclusively on your behalf, under your direction
and control, on terms that prohibit them from making any copies of
your copyrighted material outside their relationship with you.
Conveying under any other circumstances is permitted solely under
the conditions stated below. Sublicensing is not allowed; section 10
makes it unnecessary.
3. Protecting Users' Legal Rights From Anti-Circumvention Law.
No covered work shall be deemed part of an effective technological
measure under any applicable law fulfilling obligations under article
11 of the WIPO copyright treaty adopted on 20 December 1996, or
similar laws prohibiting or restricting circumvention of such
measures.
When you convey a covered work, you waive any legal power to forbid
circumvention of technological measures to the extent such circumvention
is effected by exercising rights under this License with respect to
the covered work, and you disclaim any intention to limit operation or
modification of the work as a means of enforcing, against the work's
users, your or third parties' legal rights to forbid circumvention of
technological measures.
4. Conveying Verbatim Copies.
You may convey verbatim copies of the Program's source code as you
receive it, in any medium, provided that you conspicuously and
appropriately publish on each copy an appropriate copyright notice;
keep intact all notices stating that this License and any
non-permissive terms added in accord with section 7 apply to the code;
keep intact all notices of the absence of any warranty; and give all
recipients a copy of this License along with the Program.
You may charge any price or no price for each copy that you convey,
and you may offer support or warranty protection for a fee.
5. Conveying Modified Source Versions.
You may convey a work based on the Program, or the modifications to
produce it from the Program, in the form of source code under the
terms of section 4, provided that you also meet all of these conditions:
a) The work must carry prominent notices stating that you modified
it, and giving a relevant date.
b) The work must carry prominent notices stating that it is
released under this License and any conditions added under section
7. This requirement modifies the requirement in section 4 to
"keep intact all notices".
c) You must license the entire work, as a whole, under this
License to anyone who comes into possession of a copy. This
License will therefore apply, along with any applicable section 7
additional terms, to the whole of the work, and all its parts,
regardless of how they are packaged. This License gives no
permission to license the work in any other way, but it does not
invalidate such permission if you have separately received it.
d) If the work has interactive user interfaces, each must display
Appropriate Legal Notices; however, if the Program has interactive
interfaces that do not display Appropriate Legal Notices, your
work need not make them do so.
A compilation of a covered work with other separate and independent
works, which are not by their nature extensions of the covered work,
and which are not combined with it such as to form a larger program,
in or on a volume of a storage or distribution medium, is called an
"aggregate" if the compilation and its resulting copyright are not
used to limit the access or legal rights of the compilation's users
beyond what the individual works permit. Inclusion of a covered work
in an aggregate does not cause this License to apply to the other
parts of the aggregate.
6. Conveying Non-Source Forms.
You may convey a covered work in object code form under the terms
of sections 4 and 5, provided that you also convey the
machine-readable Corresponding Source under the terms of this License,
in one of these ways:
a) Convey the object code in, or embodied in, a physical product
(including a physical distribution medium), accompanied by the
Corresponding Source fixed on a durable physical medium
customarily used for software interchange.
b) Convey the object code in, or embodied in, a physical product
(including a physical distribution medium), accompanied by a
written offer, valid for at least three years and valid for as
long as you offer spare parts or customer support for that product
model, to give anyone who possesses the object code either (1) a
copy of the Corresponding Source for all the software in the
product that is covered by this License, on a durable physical
medium customarily used for software interchange, for a price no
more than your reasonable cost of physically performing this
conveying of source, or (2) access to copy the
Corresponding Source from a network server at no charge.
c) Convey individual copies of the object code with a copy of the
written offer to provide the Corresponding Source. This
alternative is allowed only occasionally and noncommercially, and
only if you received the object code with such an offer, in accord
with subsection 6b.
d) Convey the object code by offering access from a designated
place (gratis or for a charge), and offer equivalent access to the
Corresponding Source in the same way through the same place at no
further charge. You need not require recipients to copy the
Corresponding Source along with the object code. If the place to
copy the object code is a network server, the Corresponding Source
may be on a different server (operated by you or a third party)
that supports equivalent copying facilities, provided you maintain
clear directions next to the object code saying where to find the
Corresponding Source. Regardless of what server hosts the
Corresponding Source, you remain obligated to ensure that it is
available for as long as needed to satisfy these requirements.
e) Convey the object code using peer-to-peer transmission, provided
you inform other peers where the object code and Corresponding
Source of the work are being offered to the general public at no
charge under subsection 6d.
A separable portion of the object code, whose source code is excluded
from the Corresponding Source as a System Library, need not be
included in conveying the object code work.
A "User Product" is either (1) a "consumer product", which means any
tangible personal property which is normally used for personal, family,
or household purposes, or (2) anything designed or sold for incorporation
into a dwelling. In determining whether a product is a consumer product,
doubtful cases shall be resolved in favor of coverage. For a particular
product received by a particular user, "normally used" refers to a
typical or common use of that class of product, regardless of the status
of the particular user or of the way in which the particular user
actually uses, or expects or is expected to use, the product. A product
is a consumer product regardless of whether the product has substantial
commercial, industrial or non-consumer uses, unless such uses represent
the only significant mode of use of the product.
"Installation Information" for a User Product means any methods,
procedures, authorization keys, or other information required to install
and execute modified versions of a covered work in that User Product from
a modified version of its Corresponding Source. The information must
suffice to ensure that the continued functioning of the modified object
code is in no case prevented or interfered with solely because
modification has been made.
If you convey an object code work under this section in, or with, or
specifically for use in, a User Product, and the conveying occurs as
part of a transaction in which the right of possession and use of the
User Product is transferred to the recipient in perpetuity or for a
fixed term (regardless of how the transaction is characterized), the
Corresponding Source conveyed under this section must be accompanied
by the Installation Information. But this requirement does not apply
if neither you nor any third party retains the ability to install
modified object code on the User Product (for example, the work has
been installed in ROM).
The requirement to provide Installation Information does not include a
requirement to continue to provide support service, warranty, or updates
for a work that has been modified or installed by the recipient, or for
the User Product in which it has been modified or installed. Access to a
network may be denied when the modification itself materially and
adversely affects the operation of the network or violates the rules and
protocols for communication across the network.
Corresponding Source conveyed, and Installation Information provided,
in accord with this section must be in a format that is publicly
documented (and with an implementation available to the public in
source code form), and must require no special password or key for
unpacking, reading or copying.
7. Additional Terms.
"Additional permissions" are terms that supplement the terms of this
License by making exceptions from one or more of its conditions.
Additional permissions that are applicable to the entire Program shall
be treated as though they were included in this License, to the extent
that they are valid under applicable law. If additional permissions
apply only to part of the Program, that part may be used separately
under those permissions, but the entire Program remains governed by
this License without regard to the additional permissions.
When you convey a copy of a covered work, you may at your option
remove any additional permissions from that copy, or from any part of
it. (Additional permissions may be written to require their own
removal in certain cases when you modify the work.) You may place
additional permissions on material, added by you to a covered work,
for which you have or can give appropriate copyright permission.
Notwithstanding any other provision of this License, for material you
add to a covered work, you may (if authorized by the copyright holders of
that material) supplement the terms of this License with terms:
a) Disclaiming warranty or limiting liability differently from the
terms of sections 15 and 16 of this License; or
b) Requiring preservation of specified reasonable legal notices or
author attributions in that material or in the Appropriate Legal
Notices displayed by works containing it; or
c) Prohibiting misrepresentation of the origin of that material, or
requiring that modified versions of such material be marked in
reasonable ways as different from the original version; or
d) Limiting the use for publicity purposes of names of licensors or
authors of the material; or
e) Declining to grant rights under trademark law for use of some
trade names, trademarks, or service marks; or
f) Requiring indemnification of licensors and authors of that
material by anyone who conveys the material (or modified versions of
it) with contractual assumptions of liability to the recipient, for
any liability that these contractual assumptions directly impose on
those licensors and authors.
All other non-permissive additional terms are considered "further
restrictions" within the meaning of section 10. If the Program as you
received it, or any part of it, contains a notice stating that it is
governed by this License along with a term that is a further
restriction, you may remove that term. If a license document contains
a further restriction but permits relicensing or conveying under this
License, you may add to a covered work material governed by the terms
of that license document, provided that the further restriction does
not survive such relicensing or conveying.
If you add terms to a covered work in accord with this section, you
must place, in the relevant source files, a statement of the
additional terms that apply to those files, or a notice indicating
where to find the applicable terms.
Additional terms, permissive or non-permissive, may be stated in the
form of a separately written license, or stated as exceptions;
the above requirements apply either way.
8. Termination.
You may not propagate or modify a covered work except as expressly
provided under this License. Any attempt otherwise to propagate or
modify it is void, and will automatically terminate your rights under
this License (including any patent licenses granted under the third
paragraph of section 11).
However, if you cease all violation of this License, then your
license from a particular copyright holder is reinstated (a)
provisionally, unless and until the copyright holder explicitly and
finally terminates your license, and (b) permanently, if the copyright
holder fails to notify you of the violation by some reasonable means
prior to 60 days after the cessation.
Moreover, your license from a particular copyright holder is
reinstated permanently if the copyright holder notifies you of the
violation by some reasonable means, this is the first time you have
received notice of violation of this License (for any work) from that
copyright holder, and you cure the violation prior to 30 days after
your receipt of the notice.
Termination of your rights under this section does not terminate the
licenses of parties who have received copies or rights from you under
this License. If your rights have been terminated and not permanently
reinstated, you do not qualify to receive new licenses for the same
material under section 10.
9. Acceptance Not Required for Having Copies.
You are not required to accept this License in order to receive or
run a copy of the Program. Ancillary propagation of a covered work
occurring solely as a consequence of using peer-to-peer transmission
to receive a copy likewise does not require acceptance. However,
nothing other than this License grants you permission to propagate or
modify any covered work. These actions infringe copyright if you do
not accept this License. Therefore, by modifying or propagating a
covered work, you indicate your acceptance of this License to do so.
10. Automatic Licensing of Downstream Recipients.
Each time you convey a covered work, the recipient automatically
receives a license from the original licensors, to run, modify and
propagate that work, subject to this License. You are not responsible
for enforcing compliance by third parties with this License.
An "entity transaction" is a transaction transferring control of an
organization, or substantially all assets of one, or subdividing an
organization, or merging organizations. If propagation of a covered
work results from an entity transaction, each party to that
transaction who receives a copy of the work also receives whatever
licenses to the work the party's predecessor in interest had or could
give under the previous paragraph, plus a right to possession of the
Corresponding Source of the work from the predecessor in interest, if
the predecessor has it or can get it with reasonable efforts.
You may not impose any further restrictions on the exercise of the
rights granted or affirmed under this License. For example, you may
not impose a license fee, royalty, or other charge for exercise of
rights granted under this License, and you may not initiate litigation
(including a cross-claim or counterclaim in a lawsuit) alleging that
any patent claim is infringed by making, using, selling, offering for
sale, or importing the Program or any portion of it.
11. Patents.
A "contributor" is a copyright holder who authorizes use under this
License of the Program or a work on which the Program is based. The
work thus licensed is called the contributor's "contributor version".
A contributor's "essential patent claims" are all patent claims
owned or controlled by the contributor, whether already acquired or
hereafter acquired, that would be infringed by some manner, permitted
by this License, of making, using, or selling its contributor version,
but do not include claims that would be infringed only as a
consequence of further modification of the contributor version. For
purposes of this definition, "control" includes the right to grant
patent sublicenses in a manner consistent with the requirements of
this License.
Each contributor grants you a non-exclusive, worldwide, royalty-free
patent license under the contributor's essential patent claims, to
make, use, sell, offer for sale, import and otherwise run, modify and
propagate the contents of its contributor version.
In the following three paragraphs, a "patent license" is any express
agreement or commitment, however denominated, not to enforce a patent
(such as an express permission to practice a patent or covenant not to
sue for patent infringement). To "grant" such a patent license to a
party means to make such an agreement or commitment not to enforce a
patent against the party.
If you convey a covered work, knowingly relying on a patent license,
and the Corresponding Source of the work is not available for anyone
to copy, free of charge and under the terms of this License, through a
publicly available network server or other readily accessible means,
then you must either (1) cause the Corresponding Source to be so
available, or (2) arrange to deprive yourself of the benefit of the
patent license for this particular work, or (3) arrange, in a manner
consistent with the requirements of this License, to extend the patent
license to downstream recipients. "Knowingly relying" means you have
actual knowledge that, but for the patent license, your conveying the
covered work in a country, or your recipient's use of the covered work
in a country, would infringe one or more identifiable patents in that
country that you have reason to believe are valid.
If, pursuant to or in connection with a single transaction or
arrangement, you convey, or propagate by procuring conveyance of, a
covered work, and grant a patent license to some of the parties
receiving the covered work authorizing them to use, propagate, modify
or convey a specific copy of the covered work, then the patent license
you grant is automatically extended to all recipients of the covered
work and works based on it.
A patent license is "discriminatory" if it does not include within
the scope of its coverage, prohibits the exercise of, or is
conditioned on the non-exercise of one or more of the rights that are
specifically granted under this License. You may not convey a covered
work if you are a party to an arrangement with a third party that is
in the business of distributing software, under which you make payment
to the third party based on the extent of your activity of conveying
the work, and under which the third party grants, to any of the
parties who would receive the covered work from you, a discriminatory
patent license (a) in connection with copies of the covered work
conveyed by you (or copies made from those copies), or (b) primarily
for and in connection with specific products or compilations that
contain the covered work, unless you entered into that arrangement,
or that patent license was granted, prior to 28 March 2007.
Nothing in this License shall be construed as excluding or limiting
any implied license or other defenses to infringement that may
otherwise be available to you under applicable patent law.
12. No Surrender of Others' Freedom.
If conditions are imposed on you (whether by court order, agreement or
otherwise) that contradict the conditions of this License, they do not
excuse you from the conditions of this License. If you cannot convey a
covered work so as to satisfy simultaneously your obligations under this
License and any other pertinent obligations, then as a consequence you may
not convey it at all. For example, if you agree to terms that obligate you
to collect a royalty for further conveying from those to whom you convey
the Program, the only way you could satisfy both those terms and this
License would be to refrain entirely from conveying the Program.
13. Use with the GNU Affero General Public License.
Notwithstanding any other provision of this License, you have
permission to link or combine any covered work with a work licensed
under version 3 of the GNU Affero General Public License into a single
combined work, and to convey the resulting work. The terms of this
License will continue to apply to the part which is the covered work,
but the special requirements of the GNU Affero General Public License,
section 13, concerning interaction through a network will apply to the
combination as such.
14. Revised Versions of this License.
The Free Software Foundation may publish revised and/or new versions of
the GNU General Public License from time to time. Such new versions will
be similar in spirit to the present version, but may differ in detail to
address new problems or concerns.
Each version is given a distinguishing version number. If the
Program specifies that a certain numbered version of the GNU General
Public License "or any later version" applies to it, you have the
option of following the terms and conditions either of that numbered
version or of any later version published by the Free Software
Foundation. If the Program does not specify a version number of the
GNU General Public License, you may choose any version ever published
by the Free Software Foundation.
If the Program specifies that a proxy can decide which future
versions of the GNU General Public License can be used, that proxy's
public statement of acceptance of a version permanently authorizes you
to choose that version for the Program.
Later license versions may give you additional or different
permissions. However, no additional obligations are imposed on any
author or copyright holder as a result of your choosing to follow a
later version.
15. Disclaimer of Warranty.
THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
16. Limitation of Liability.
IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
SUCH DAMAGES.
17. Interpretation of Sections 15 and 16.
If the disclaimer of warranty and limitation of liability provided
above cannot be given local legal effect according to their terms,
reviewing courts shall apply local law that most closely approximates
an absolute waiver of all civil liability in connection with the
Program, unless a warranty or assumption of liability accompanies a
copy of the Program in return for a fee.
END OF TERMS AND CONDITIONS
How to Apply These Terms to Your New Programs
If you develop a new program, and you want it to be of the greatest
possible use to the public, the best way to achieve this is to make it
free software which everyone can redistribute and change under these terms.
To do so, attach the following notices to the program. It is safest
to attach them to the start of each source file to most effectively
state the exclusion of warranty; and each file should have at least
the "copyright" line and a pointer to where the full notice is found.
<one line to give the program's name and a brief idea of what it does.>
Copyright (C) <year> <name of author>
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <https://www.gnu.org/licenses/>.
Also add information on how to contact you by electronic and paper mail.
If the program does terminal interaction, make it output a short
notice like this when it starts in an interactive mode:
<program> Copyright (C) <year> <name of author>
This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
This is free software, and you are welcome to redistribute it
under certain conditions; type `show c' for details.
The hypothetical commands `show w' and `show c' should show the appropriate
parts of the General Public License. Of course, your program's commands
might be different; for a GUI interface, you would use an "about box".
You should also get your employer (if you work as a programmer) or school,
if any, to sign a "copyright disclaimer" for the program, if necessary.
For more information on this, and how to apply and follow the GNU GPL, see
<https://www.gnu.org/licenses/>.
The GNU General Public License does not permit incorporating your program
into proprietary programs. If your program is a subroutine library, you
may consider it more useful to permit linking proprietary applications with
the library. If this is what you want to do, use the GNU Lesser General
Public License instead of this License. But first, please read
<https://www.gnu.org/licenses/why-not-lgpl.html>.

42
README.md Normal file
View file

@ -0,0 +1,42 @@
# flutter_stockfish_plugin
A Flutter plugin for Stockfish Chess engine.
The current Version is based on Stockfish 16.
## Usage
```dart
final stockfish = new Stockfish()
// Listen on stdout of Stockfish engine
final stockfishSubscription = stockfish.stdout.listen((line) {
print("received: $line");
});
// Sending UCI command to get Stockfish ready
stockfish.stdin = 'isready'
stockfish.stdin = 'position startpos moves e2e4' // set up start position
stockfish.stdin = 'go depth 20' // search bestmove with a max septh of 20
// Don't remember to dispose Stockfish when you're done.
// Make shure to dispose Stockfish when closing the app. May use WindowListener.
stockfishSubscription.cancel();
stockfish.dispose();
```
## Goal of this fork of stockfish_chess_engine
* Avoid limitation. This Version does not redirect stdout and stdin of the app for communication with stockfish.
* stdin and stdout where replaced with a fakestream element.
* Stockfish internel Logging might not work (could be fixed).
## Credits
* Based on and using source code from [stockfish_chess_engine](https://github.com/loloof64/StockfishChessEngineFlutter)
* Using source code from [Stockfish](https://stockfishchess.org).
* Using source code from [Flutter Stockfish](https://github.com/ArjanAswal/Stockfish).
Directory src/Stockfish contains the latest current release.
The code is modified to use a different communication interface.
Original License of [Stockfish](https://stockfishchess.org) can be found in their [Github](https://github.com/official-stockfish/Stockfish) Repository.

4
analysis_options.yaml Normal file
View file

@ -0,0 +1,4 @@
include: package:flutter_lints/flutter.yaml
# Additional information about this file can be found at
# https://dart.dev/guides/language/analysis-options

9
android/.gitignore vendored Normal file
View file

@ -0,0 +1,9 @@
*.iml
.gradle
/local.properties
/.idea/workspace.xml
/.idea/libraries
.DS_Store
/build
/captures
.cxx

59
android/build.gradle Normal file
View file

@ -0,0 +1,59 @@
// The Android Gradle Plugin builds the native code with the Android NDK.
group 'de.jusax.stockfish'
version '1.0'
buildscript {
repositories {
google()
mavenCentral()
}
dependencies {
// The Android Gradle Plugin knows how to build native code with the NDK.
classpath 'com.android.tools.build:gradle:7.1.2'
}
}
rootProject.allprojects {
repositories {
google()
mavenCentral()
}
}
apply plugin: 'com.android.library'
android {
// Bumping the plugin compileSdkVersion requires all clients of this plugin
// to bump the version in their app.
compileSdkVersion 31
// Bumping the plugin ndkVersion requires all clients of this plugin to bump
// the version in their app and to download a newer version of the NDK.
ndkVersion "21.1.6352462"
// Invoke the shared CMake build with the Android Gradle Plugin.
externalNativeBuild {
cmake {
path "../src/CMakeLists.txt"
// The default CMake version for the Android Gradle Plugin is 3.10.2.
// https://developer.android.com/studio/projects/install-ndk#vanilla_cmake
//
// The Flutter tooling requires that developers have CMake 3.10 or later
// installed. You should not increase this version, as doing so will cause
// the plugin to fail to compile for some customers of the plugin.
// version "3.10.2"
}
}
compileOptions {
sourceCompatibility JavaVersion.VERSION_1_8
targetCompatibility JavaVersion.VERSION_1_8
}
defaultConfig {
minSdkVersion 16
}
}

1
android/settings.gradle Normal file
View file

@ -0,0 +1 @@
rootProject.name = 'stockfish'

3
android/src/main/AndroidManifest.xml Normal file
View file

@ -0,0 +1,3 @@
<manifest xmlns:android="http://schemas.android.com/apk/res/android"
package="de.jusax.stockfish">
</manifest>

21
ffigen.yaml Normal file
View file

@ -0,0 +1,21 @@
# Run with `flutter pub run ffigen --config ffigen_linux.yaml`.
name: StockfishChessEngineBindings
description: |
Bindings for `src/stockfish.h`.
Regenerate bindings with `dart run ffigen --config ffigen.yaml`.
output: 'lib/stockfish_bindings_generated.dart'
headers:
entry-points:
- 'src/stockfish.h'
include-directives:
- 'src/stockfish.h'
preamble: |
// ignore_for_file: always_specify_types
// ignore_for_file: camel_case_types
// ignore_for_file: non_constant_identifier_names
comments:
style: any
length: full
llvm-path:
- /usr/lib/llvm-14

1
lib/stockfish.dart
View file

@ -165,7 +165,6 @@ void _isolateMain(SendPort mainPort) {
mainPort.send(exitCode);
developer.log('nativeMain returns $exitCode', name: 'Stockfish');
print("stoped");
}
void _isolateStdout(SendPort stdoutPort) {

76
pubspec.yaml Normal file
View file

@ -0,0 +1,76 @@
name: flutter_stockfish_plugin
description: "A Stockfish Plugin for Flutter."
version: 1.0.0
homepage:
environment:
sdk: '>=3.2.0-210.3.beta <4.0.0'
flutter: '>=3.3.0'
dependencies:
ffi: ^2.1.0
flutter:
sdk: flutter
plugin_platform_interface: ^2.0.2
dev_dependencies:
flutter_test:
sdk: flutter
flutter_lints: ^2.0.0
ffigen: ^9.0.1
# For information on the generic Dart part of this file, see the
# following page: https://dart.dev/tools/pub/pubspec
# The following section is specific to Flutter packages.
flutter:
# This section identifies this Flutter project as a plugin project.
# The 'pluginClass' specifies the class (in Java, Kotlin, Swift, Objective-C, etc.)
# which should be registered in the plugin registry. This is required for
# using method channels.
# The Android 'package' specifies package in which the registered class is.
# This is required for using method channels on Android.
# The 'ffiPlugin' specifies that native code should be built and bundled.
# This is required for using `dart:ffi`.
# All these are used by the tooling to maintain consistency when
# adding or updating assets for this project.
plugin:
platforms:
android:
ffiPlugin: true
linux:
ffiPlugin: true
windows:
ffiPlugin: true
# -------------------
# To add assets to your plugin package, add an assets section, like this:
# assets:
# - images/a_dot_burr.jpeg
# - images/a_dot_ham.jpeg
#
# For details regarding assets in packages, see
# https://flutter.dev/assets-and-images/#from-packages
#
# An image asset can refer to one or more resolution-specific "variants", see
# https://flutter.dev/assets-and-images/#resolution-aware
# To add custom fonts to your plugin package, add a fonts section here,
# in this "flutter" section. Each entry in this list should have a
# "family" key with the font family name, and a "fonts" key with a
# list giving the asset and other descriptors for the font. For
# example:
# fonts:
# - family: Schyler
# fonts:
# - asset: fonts/Schyler-Regular.ttf
# - asset: fonts/Schyler-Italic.ttf
# style: italic
# - family: Trajan Pro
# fonts:
# - asset: fonts/TrajanPro.ttf
# - asset: fonts/TrajanPro_Bold.ttf
# weight: 700
#
# For details regarding fonts in packages, see
# https://flutter.dev/custom-fonts/#from-packages

5
src/CMakeLists.txt
View file

@ -6,7 +6,10 @@ cmake_minimum_required(VERSION 3.18)
project(stockfish VERSION 0.0.1 LANGUAGES CXX)
file(GLOB_RECURSE cppPaths "Stockfish/src/*.cpp")
set(CMAKE_CXX_STANDARD 17)
set(NNUE_NAME nn-0000000000a0.nnue)
set(CMAKE_CXX_FLAGS "-O3 -flto -march=native")
#old: -Wshadow -Wmissing-declarations -fno-exceptions -std=c++17 -fPIC -Wall -Wextra -march=native
#set(CMAKE_CXX_FLAGS_RELEASE "-O3 -flto -march=native")
set(NNUE_NAME nn-5af11540bbfe.nnue)
add_library(stockfish SHARED
"stockfish.cpp"

12
src/Stockfish/.gitignore vendored Normal file
View file

@ -0,0 +1,12 @@
# Files from build
**/*.o
**/*.s
src/.depend
# Built binary
src/stockfish*
src/-lstdc++.res
# Neural network for the NNUE evaluation
**/*.nnue

226
src/Stockfish/AUTHORS Normal file
View file

@ -0,0 +1,226 @@
# Founders of the Stockfish project and Fishtest infrastructure
Tord Romstad (romstad)
Marco Costalba (mcostalba)
Joona Kiiski (zamar)
Gary Linscott (glinscott)
# Authors and inventors of NNUE, training, and NNUE port
Yu Nasu (ynasu87)
Motohiro Isozaki (yaneurao)
Hisayori Noda (nodchip)
# All other authors of Stockfish code (in alphabetical order)
Aditya (absimaldata)
Adrian Petrescu (apetresc)
Ajith Chandy Jose (ajithcj)
Alain Savard (Rocky640)
Alayan Feh (Alayan-stk-2)
Alexander Kure
Alexander Pagel (Lolligerhans)
Alfredo Menezes (lonfom169)
Ali AlZhrani (Cooffe)
Andreas Matthies (Matthies)
Andrei Vetrov (proukornew)
Andrew Grant (AndyGrant)
Andrey Neporada (nepal)
Andy Duplain
Antoine Champion (antoinechampion)
Aram Tumanian (atumanian)
Arjun Temurnikar
Artem Solopiy (EntityFX)
Auguste Pop
Balint Pfliegel
Ben Chaney (Chaneybenjamini)
Ben Koshy (BKSpurgeon)
Bill Henry (VoyagerOne)
Bojun Guo (noobpwnftw, Nooby)
borg323
Boštjan Mejak (PedanticHacker)
braich
Brian Sheppard (SapphireBrand, briansheppard-toast)
Bruno de Melo Costa (BM123499)
Bruno Pellanda (pellanda)
Bryan Cross (crossbr)
candirufish
Chess13234
Chris Cain (ceebo)
clefrks
Dale Weiler (graphitemaster)
Daniel Axtens (daxtens)
Daniel Dugovic (ddugovic)
Dan Schmidt (dfannius)
Dariusz Orzechowski (dorzechowski)
David (dav1312)
David Zar
Daylen Yang (daylen)
Deshawn Mohan-Smith (GoldenRare)
Dieter Dobbelaere (ddobbelaere)
DiscanX
Dominik Schlösser (domschl)
double-beep
Douglas Matos Gomes (dsmsgms)
Dubslow
Eduardo Cáceres (eduherminio)
Eelco de Groot (KingDefender)
Elvin Liu (solarlight2)
erbsenzaehler
Ernesto Gatti
Fabian Beuke (madnight)
Fabian Fichter (ianfab)
Fanael Linithien (Fanael)
fanon
Fauzi Akram Dabat (FauziAkram)
Felix Wittmann
gamander
Gary Heckman (gheckman)
George Sobala (gsobala)
gguliash
Giacomo Lorenzetti (G-Lorenz)
Gian-Carlo Pascutto (gcp)
Goh CJ (cj5716)
Gontran Lemaire (gonlem)
Goodkov Vasiliy Aleksandrovich (goodkov)
Gregor Cramer
GuardianRM
Guy Vreuls (gvreuls)
Günther Demetz (pb00067, pb00068)
Henri Wiechers
Hiraoka Takuya (HiraokaTakuya)
homoSapiensSapiens
Hongzhi Cheng
Ivan Ivec (IIvec)
Jacques B. (Timshel)
Jake Senne (w1wwwwww)
Jan Ondruš (hxim)
Jared Kish (Kurtbusch, kurt22i)
Jarrod Torriero (DU-jdto)
Jean-Francois Romang (jromang)
Jean Gauthier (OuaisBla)
Jekaa
Jerry Donald Watson (jerrydonaldwatson)
jjoshua2
Jonathan Buladas Dumale (SFisGOD)
Jonathan Calovski (Mysseno)
Jonathan McDermid (jonathanmcdermid)
Joost VandeVondele (vondele)
Joseph Ellis (jhellis3)
Joseph R. Prostko
Jörg Oster (joergoster)
Julian Willemer (NightlyKing)
jundery
Justin Blanchard (UncombedCoconut)
Kelly Wilson
Ken Takusagawa
Kian E (KJE-98)
kinderchocolate
Kiran Panditrao (Krgp)
Kojirion
Krystian Kuzniarek (kuzkry)
Leonardo Ljubičić (ICCF World Champion)
Leonid Pechenik (lp--)
Liam Keegan (lkeegan)
Linmiao Xu (linrock)
Linus Arver (listx)
loco-loco
Lub van den Berg (ElbertoOne)
Luca Brivio (lucabrivio)
Lucas Braesch (lucasart)
Lyudmil Antonov (lantonov)
Maciej Żenczykowski (zenczykowski)
Malcolm Campbell (xoto10)
Mark Tenzer (31m059)
marotear
Matt Ginsberg (mattginsberg)
Matthew Lai (matthewlai)
Matthew Sullivan (Matt14916)
Max A. (Disservin)
Maxim Masiutin (maximmasiutin)
Maxim Molchanov (Maxim)
Michael An (man)
Michael Byrne (MichaelB7)
Michael Chaly (Vizvezdenec)
Michael Stembera (mstembera)
Michael Whiteley (protonspring)
Michel Van den Bergh (vdbergh)
Miguel Lahoz (miguel-l)
Mikael Bäckman (mbootsector)
Mike Babigian (Farseer)
Mira
Miroslav Fontán (Hexik)
Moez Jellouli (MJZ1977)
Mohammed Li (tthsqe12)
Muzhen J (XInTheDark)
Nathan Rugg (nmrugg)
Nguyen Pham (nguyenpham)
Nicklas Persson (NicklasPersson)
Nick Pelling (nickpelling)
Niklas Fiekas (niklasf)
Nikolay Kostov (NikolayIT)
Norman Schmidt (FireFather)
notruck
Ofek Shochat (OfekShochat, ghostway)
Ondrej Mosnáček (WOnder93)
Ondřej Mišina (AndrovT)
Oskar Werkelin Ahlin
Pablo Vazquez
Panthee
Pascal Romaret
Pasquale Pigazzini (ppigazzini)
Patrick Jansen (mibere)
Peter Schneider (pschneider1968)
Peter Zsifkovits (CoffeeOne)
PikaCat
Praveen Kumar Tummala (praveentml)
Prokop Randáček (ProkopRandacek)
Rahul Dsilva (silversolver1)
Ralph Stößer (Ralph Stoesser)
Raminder Singh
renouve
Reuven Peleg (R-Peleg)
Richard Lloyd (Richard-Lloyd)
Rodrigo Exterckötter Tjäder
Rodrigo Roim (roim)
Ronald de Man (syzygy1, syzygy)
Ron Britvich (Britvich)
rqs
Rui Coelho (ruicoelhopedro)
Ryan Schmitt
Ryan Takker
Sami Kiminki (skiminki)
Sebastian Buchwald (UniQP)
Sergei Antonov (saproj)
Sergei Ivanov (svivanov72)
Sergio Vieri (sergiovieri)
sf-x
Shahin M. Shahin (peregrine)
Shane Booth (shane31)
Shawn Varghese (xXH4CKST3RXx)
Siad Daboul (Topologist)
Stefan Geschwentner (locutus2)
Stefano Cardanobile (Stefano80)
Stefano Di Martino (StefanoD)
Steinar Gunderson (sesse)
Stéphane Nicolet (snicolet)
Syine Mineta (MinetaS)
Thanar2
thaspel
theo77186
Tomasz Sobczyk (Sopel97)
Tom Truscott
Tom Vijlbrief (tomtor)
Torsten Franz (torfranz, tfranzer)
Torsten Hellwig (Torom)
Tracey Emery (basepr1me)
tttak
Unai Corzo (unaiic)
Uri Blass (uriblass)
Vince Negri (cuddlestmonkey)
Viren
windfishballad
xefoci7612
zz4032
# Additionally, we acknowledge the authors and maintainers of fishtest,
# an amazing and essential framework for Stockfish development!
#
# https://github.com/glinscott/fishtest/blob/master/AUTHORS

23
src/Stockfish/CITATION.cff Normal file
View file

@ -0,0 +1,23 @@
# This CITATION.cff file was generated with cffinit.
# Visit https://bit.ly/cffinit to generate yours today!
cff-version: 1.2.0
title: Stockfish
message: >-
Please cite this software using the metadata from this
file.
type: software
authors:
- name: The Stockfish developers (see AUTHORS file)
repository-code: 'https://github.com/official-stockfish/Stockfish'
url: 'https://stockfishchess.org/'
repository-artifact: 'https://stockfishchess.org/download/'
abstract: Stockfish is a free and strong UCI chess engine.
keywords:
- chess
- artificial intelligence (AI)
- tree search
- alpha-beta search
- neural networks (NN)
- efficiently updatable neural networks (NNUE)
license: GPL-3.0

674
src/Stockfish/Copying.txt Normal file
View file

@ -0,0 +1,674 @@
GNU GENERAL PUBLIC LICENSE
Version 3, 29 June 2007
Copyright (C) 2007 Free Software Foundation, Inc. <http://fsf.org/>
Everyone is permitted to copy and distribute verbatim copies
of this license document, but changing it is not allowed.
Preamble
The GNU General Public License is a free, copyleft license for
software and other kinds of works.
The licenses for most software and other practical works are designed
to take away your freedom to share and change the works. By contrast,
the GNU General Public License is intended to guarantee your freedom to
share and change all versions of a program--to make sure it remains free
software for all its users. We, the Free Software Foundation, use the
GNU General Public License for most of our software; it applies also to
any other work released this way by its authors. You can apply it to
your programs, too.
When we speak of free software, we are referring to freedom, not
price. Our General Public Licenses are designed to make sure that you
have the freedom to distribute copies of free software (and charge for
them if you wish), that you receive source code or can get it if you
want it, that you can change the software or use pieces of it in new
free programs, and that you know you can do these things.
To protect your rights, we need to prevent others from denying you
these rights or asking you to surrender the rights. Therefore, you have
certain responsibilities if you distribute copies of the software, or if
you modify it: responsibilities to respect the freedom of others.
For example, if you distribute copies of such a program, whether
gratis or for a fee, you must pass on to the recipients the same
freedoms that you received. You must make sure that they, too, receive
or can get the source code. And you must show them these terms so they
know their rights.
Developers that use the GNU GPL protect your rights with two steps:
(1) assert copyright on the software, and (2) offer you this License
giving you legal permission to copy, distribute and/or modify it.
For the developers' and authors' protection, the GPL clearly explains
that there is no warranty for this free software. For both users' and
authors' sake, the GPL requires that modified versions be marked as
changed, so that their problems will not be attributed erroneously to
authors of previous versions.
Some devices are designed to deny users access to install or run
modified versions of the software inside them, although the manufacturer
can do so. This is fundamentally incompatible with the aim of
protecting users' freedom to change the software. The systematic
pattern of such abuse occurs in the area of products for individuals to
use, which is precisely where it is most unacceptable. Therefore, we
have designed this version of the GPL to prohibit the practice for those
products. If such problems arise substantially in other domains, we
stand ready to extend this provision to those domains in future versions
of the GPL, as needed to protect the freedom of users.
Finally, every program is threatened constantly by software patents.
States should not allow patents to restrict development and use of
software on general-purpose computers, but in those that do, we wish to
avoid the special danger that patents applied to a free program could
make it effectively proprietary. To prevent this, the GPL assures that
patents cannot be used to render the program non-free.
The precise terms and conditions for copying, distribution and
modification follow.
TERMS AND CONDITIONS
0. Definitions.
"This License" refers to version 3 of the GNU General Public License.
"Copyright" also means copyright-like laws that apply to other kinds of
works, such as semiconductor masks.
"The Program" refers to any copyrightable work licensed under this
License. Each licensee is addressed as "you". "Licensees" and
"recipients" may be individuals or organizations.
To "modify" a work means to copy from or adapt all or part of the work
in a fashion requiring copyright permission, other than the making of an
exact copy. The resulting work is called a "modified version" of the
earlier work or a work "based on" the earlier work.
A "covered work" means either the unmodified Program or a work based
on the Program.
To "propagate" a work means to do anything with it that, without
permission, would make you directly or secondarily liable for
infringement under applicable copyright law, except executing it on a
computer or modifying a private copy. Propagation includes copying,
distribution (with or without modification), making available to the
public, and in some countries other activities as well.
To "convey" a work means any kind of propagation that enables other
parties to make or receive copies. Mere interaction with a user through
a computer network, with no transfer of a copy, is not conveying.
An interactive user interface displays "Appropriate Legal Notices"
to the extent that it includes a convenient and prominently visible
feature that (1) displays an appropriate copyright notice, and (2)
tells the user that there is no warranty for the work (except to the
extent that warranties are provided), that licensees may convey the
work under this License, and how to view a copy of this License. If
the interface presents a list of user commands or options, such as a
menu, a prominent item in the list meets this criterion.
1. Source Code.
The "source code" for a work means the preferred form of the work
for making modifications to it. "Object code" means any non-source
form of a work.
A "Standard Interface" means an interface that either is an official
standard defined by a recognized standards body, or, in the case of
interfaces specified for a particular programming language, one that
is widely used among developers working in that language.
The "System Libraries" of an executable work include anything, other
than the work as a whole, that (a) is included in the normal form of
packaging a Major Component, but which is not part of that Major
Component, and (b) serves only to enable use of the work with that
Major Component, or to implement a Standard Interface for which an
implementation is available to the public in source code form. A
"Major Component", in this context, means a major essential component
(kernel, window system, and so on) of the specific operating system
(if any) on which the executable work runs, or a compiler used to
produce the work, or an object code interpreter used to run it.
The "Corresponding Source" for a work in object code form means all
the source code needed to generate, install, and (for an executable
work) run the object code and to modify the work, including scripts to
control those activities. However, it does not include the work's
System Libraries, or general-purpose tools or generally available free
programs which are used unmodified in performing those activities but
which are not part of the work. For example, Corresponding Source
includes interface definition files associated with source files for
the work, and the source code for shared libraries and dynamically
linked subprograms that the work is specifically designed to require,
such as by intimate data communication or control flow between those
subprograms and other parts of the work.
The Corresponding Source need not include anything that users
can regenerate automatically from other parts of the Corresponding
Source.
The Corresponding Source for a work in source code form is that
same work.
2. Basic Permissions.
All rights granted under this License are granted for the term of
copyright on the Program, and are irrevocable provided the stated
conditions are met. This License explicitly affirms your unlimited
permission to run the unmodified Program. The output from running a
covered work is covered by this License only if the output, given its
content, constitutes a covered work. This License acknowledges your
rights of fair use or other equivalent, as provided by copyright law.
You may make, run and propagate covered works that you do not
convey, without conditions so long as your license otherwise remains
in force. You may convey covered works to others for the sole purpose
of having them make modifications exclusively for you, or provide you
with facilities for running those works, provided that you comply with
the terms of this License in conveying all material for which you do
not control copyright. Those thus making or running the covered works
for you must do so exclusively on your behalf, under your direction
and control, on terms that prohibit them from making any copies of
your copyrighted material outside their relationship with you.
Conveying under any other circumstances is permitted solely under
the conditions stated below. Sublicensing is not allowed; section 10
makes it unnecessary.
3. Protecting Users' Legal Rights From Anti-Circumvention Law.
No covered work shall be deemed part of an effective technological
measure under any applicable law fulfilling obligations under article
11 of the WIPO copyright treaty adopted on 20 December 1996, or
similar laws prohibiting or restricting circumvention of such
measures.
When you convey a covered work, you waive any legal power to forbid
circumvention of technological measures to the extent such circumvention
is effected by exercising rights under this License with respect to
the covered work, and you disclaim any intention to limit operation or
modification of the work as a means of enforcing, against the work's
users, your or third parties' legal rights to forbid circumvention of
technological measures.
4. Conveying Verbatim Copies.
You may convey verbatim copies of the Program's source code as you
receive it, in any medium, provided that you conspicuously and
appropriately publish on each copy an appropriate copyright notice;
keep intact all notices stating that this License and any
non-permissive terms added in accord with section 7 apply to the code;
keep intact all notices of the absence of any warranty; and give all
recipients a copy of this License along with the Program.
You may charge any price or no price for each copy that you convey,
and you may offer support or warranty protection for a fee.
5. Conveying Modified Source Versions.
You may convey a work based on the Program, or the modifications to
produce it from the Program, in the form of source code under the
terms of section 4, provided that you also meet all of these conditions:
a) The work must carry prominent notices stating that you modified
it, and giving a relevant date.
b) The work must carry prominent notices stating that it is
released under this License and any conditions added under section
7. This requirement modifies the requirement in section 4 to
"keep intact all notices".
c) You must license the entire work, as a whole, under this
License to anyone who comes into possession of a copy. This
License will therefore apply, along with any applicable section 7
additional terms, to the whole of the work, and all its parts,
regardless of how they are packaged. This License gives no
permission to license the work in any other way, but it does not
invalidate such permission if you have separately received it.
d) If the work has interactive user interfaces, each must display
Appropriate Legal Notices; however, if the Program has interactive
interfaces that do not display Appropriate Legal Notices, your
work need not make them do so.
A compilation of a covered work with other separate and independent
works, which are not by their nature extensions of the covered work,
and which are not combined with it such as to form a larger program,
in or on a volume of a storage or distribution medium, is called an
"aggregate" if the compilation and its resulting copyright are not
used to limit the access or legal rights of the compilation's users
beyond what the individual works permit. Inclusion of a covered work
in an aggregate does not cause this License to apply to the other
parts of the aggregate.
6. Conveying Non-Source Forms.
You may convey a covered work in object code form under the terms
of sections 4 and 5, provided that you also convey the
machine-readable Corresponding Source under the terms of this License,
in one of these ways:
a) Convey the object code in, or embodied in, a physical product
(including a physical distribution medium), accompanied by the
Corresponding Source fixed on a durable physical medium
customarily used for software interchange.
b) Convey the object code in, or embodied in, a physical product
(including a physical distribution medium), accompanied by a
written offer, valid for at least three years and valid for as
long as you offer spare parts or customer support for that product
model, to give anyone who possesses the object code either (1) a
copy of the Corresponding Source for all the software in the
product that is covered by this License, on a durable physical
medium customarily used for software interchange, for a price no
more than your reasonable cost of physically performing this
conveying of source, or (2) access to copy the
Corresponding Source from a network server at no charge.
c) Convey individual copies of the object code with a copy of the
written offer to provide the Corresponding Source. This
alternative is allowed only occasionally and noncommercially, and
only if you received the object code with such an offer, in accord
with subsection 6b.
d) Convey the object code by offering access from a designated
place (gratis or for a charge), and offer equivalent access to the
Corresponding Source in the same way through the same place at no
further charge. You need not require recipients to copy the
Corresponding Source along with the object code. If the place to
copy the object code is a network server, the Corresponding Source
may be on a different server (operated by you or a third party)
that supports equivalent copying facilities, provided you maintain
clear directions next to the object code saying where to find the
Corresponding Source. Regardless of what server hosts the
Corresponding Source, you remain obligated to ensure that it is
available for as long as needed to satisfy these requirements.
e) Convey the object code using peer-to-peer transmission, provided
you inform other peers where the object code and Corresponding
Source of the work are being offered to the general public at no
charge under subsection 6d.
A separable portion of the object code, whose source code is excluded
from the Corresponding Source as a System Library, need not be
included in conveying the object code work.
A "User Product" is either (1) a "consumer product", which means any
tangible personal property which is normally used for personal, family,
or household purposes, or (2) anything designed or sold for incorporation
into a dwelling. In determining whether a product is a consumer product,
doubtful cases shall be resolved in favor of coverage. For a particular
product received by a particular user, "normally used" refers to a
typical or common use of that class of product, regardless of the status
of the particular user or of the way in which the particular user
actually uses, or expects or is expected to use, the product. A product
is a consumer product regardless of whether the product has substantial
commercial, industrial or non-consumer uses, unless such uses represent
the only significant mode of use of the product.
"Installation Information" for a User Product means any methods,
procedures, authorization keys, or other information required to install
and execute modified versions of a covered work in that User Product from
a modified version of its Corresponding Source. The information must
suffice to ensure that the continued functioning of the modified object
code is in no case prevented or interfered with solely because
modification has been made.
If you convey an object code work under this section in, or with, or
specifically for use in, a User Product, and the conveying occurs as
part of a transaction in which the right of possession and use of the
User Product is transferred to the recipient in perpetuity or for a
fixed term (regardless of how the transaction is characterized), the
Corresponding Source conveyed under this section must be accompanied
by the Installation Information. But this requirement does not apply
if neither you nor any third party retains the ability to install
modified object code on the User Product (for example, the work has
been installed in ROM).
The requirement to provide Installation Information does not include a
requirement to continue to provide support service, warranty, or updates
for a work that has been modified or installed by the recipient, or for
the User Product in which it has been modified or installed. Access to a
network may be denied when the modification itself materially and
adversely affects the operation of the network or violates the rules and
protocols for communication across the network.
Corresponding Source conveyed, and Installation Information provided,
in accord with this section must be in a format that is publicly
documented (and with an implementation available to the public in
source code form), and must require no special password or key for
unpacking, reading or copying.
7. Additional Terms.
"Additional permissions" are terms that supplement the terms of this
License by making exceptions from one or more of its conditions.
Additional permissions that are applicable to the entire Program shall
be treated as though they were included in this License, to the extent
that they are valid under applicable law. If additional permissions
apply only to part of the Program, that part may be used separately
under those permissions, but the entire Program remains governed by
this License without regard to the additional permissions.
When you convey a copy of a covered work, you may at your option
remove any additional permissions from that copy, or from any part of
it. (Additional permissions may be written to require their own
removal in certain cases when you modify the work.) You may place
additional permissions on material, added by you to a covered work,
for which you have or can give appropriate copyright permission.
Notwithstanding any other provision of this License, for material you
add to a covered work, you may (if authorized by the copyright holders of
that material) supplement the terms of this License with terms:
a) Disclaiming warranty or limiting liability differently from the
terms of sections 15 and 16 of this License; or
b) Requiring preservation of specified reasonable legal notices or
author attributions in that material or in the Appropriate Legal
Notices displayed by works containing it; or
c) Prohibiting misrepresentation of the origin of that material, or
requiring that modified versions of such material be marked in
reasonable ways as different from the original version; or
d) Limiting the use for publicity purposes of names of licensors or
authors of the material; or
e) Declining to grant rights under trademark law for use of some
trade names, trademarks, or service marks; or
f) Requiring indemnification of licensors and authors of that
material by anyone who conveys the material (or modified versions of
it) with contractual assumptions of liability to the recipient, for
any liability that these contractual assumptions directly impose on
those licensors and authors.
All other non-permissive additional terms are considered "further
restrictions" within the meaning of section 10. If the Program as you
received it, or any part of it, contains a notice stating that it is
governed by this License along with a term that is a further
restriction, you may remove that term. If a license document contains
a further restriction but permits relicensing or conveying under this
License, you may add to a covered work material governed by the terms
of that license document, provided that the further restriction does
not survive such relicensing or conveying.
If you add terms to a covered work in accord with this section, you
must place, in the relevant source files, a statement of the
additional terms that apply to those files, or a notice indicating
where to find the applicable terms.
Additional terms, permissive or non-permissive, may be stated in the
form of a separately written license, or stated as exceptions;
the above requirements apply either way.
8. Termination.
You may not propagate or modify a covered work except as expressly
provided under this License. Any attempt otherwise to propagate or
modify it is void, and will automatically terminate your rights under
this License (including any patent licenses granted under the third
paragraph of section 11).
However, if you cease all violation of this License, then your
license from a particular copyright holder is reinstated (a)
provisionally, unless and until the copyright holder explicitly and
finally terminates your license, and (b) permanently, if the copyright
holder fails to notify you of the violation by some reasonable means
prior to 60 days after the cessation.
Moreover, your license from a particular copyright holder is
reinstated permanently if the copyright holder notifies you of the
violation by some reasonable means, this is the first time you have
received notice of violation of this License (for any work) from that
copyright holder, and you cure the violation prior to 30 days after
your receipt of the notice.
Termination of your rights under this section does not terminate the
licenses of parties who have received copies or rights from you under
this License. If your rights have been terminated and not permanently
reinstated, you do not qualify to receive new licenses for the same
material under section 10.
9. Acceptance Not Required for Having Copies.
You are not required to accept this License in order to receive or
run a copy of the Program. Ancillary propagation of a covered work
occurring solely as a consequence of using peer-to-peer transmission
to receive a copy likewise does not require acceptance. However,
nothing other than this License grants you permission to propagate or
modify any covered work. These actions infringe copyright if you do
not accept this License. Therefore, by modifying or propagating a
covered work, you indicate your acceptance of this License to do so.
10. Automatic Licensing of Downstream Recipients.
Each time you convey a covered work, the recipient automatically
receives a license from the original licensors, to run, modify and
propagate that work, subject to this License. You are not responsible
for enforcing compliance by third parties with this License.
An "entity transaction" is a transaction transferring control of an
organization, or substantially all assets of one, or subdividing an
organization, or merging organizations. If propagation of a covered
work results from an entity transaction, each party to that
transaction who receives a copy of the work also receives whatever
licenses to the work the party's predecessor in interest had or could
give under the previous paragraph, plus a right to possession of the
Corresponding Source of the work from the predecessor in interest, if
the predecessor has it or can get it with reasonable efforts.
You may not impose any further restrictions on the exercise of the
rights granted or affirmed under this License. For example, you may
not impose a license fee, royalty, or other charge for exercise of
rights granted under this License, and you may not initiate litigation
(including a cross-claim or counterclaim in a lawsuit) alleging that
any patent claim is infringed by making, using, selling, offering for
sale, or importing the Program or any portion of it.
11. Patents.
A "contributor" is a copyright holder who authorizes use under this
License of the Program or a work on which the Program is based. The
work thus licensed is called the contributor's "contributor version".
A contributor's "essential patent claims" are all patent claims
owned or controlled by the contributor, whether already acquired or
hereafter acquired, that would be infringed by some manner, permitted
by this License, of making, using, or selling its contributor version,
but do not include claims that would be infringed only as a
consequence of further modification of the contributor version. For
purposes of this definition, "control" includes the right to grant
patent sublicenses in a manner consistent with the requirements of
this License.
Each contributor grants you a non-exclusive, worldwide, royalty-free
patent license under the contributor's essential patent claims, to
make, use, sell, offer for sale, import and otherwise run, modify and
propagate the contents of its contributor version.
In the following three paragraphs, a "patent license" is any express
agreement or commitment, however denominated, not to enforce a patent
(such as an express permission to practice a patent or covenant not to
sue for patent infringement). To "grant" such a patent license to a
party means to make such an agreement or commitment not to enforce a
patent against the party.
If you convey a covered work, knowingly relying on a patent license,
and the Corresponding Source of the work is not available for anyone
to copy, free of charge and under the terms of this License, through a
publicly available network server or other readily accessible means,
then you must either (1) cause the Corresponding Source to be so
available, or (2) arrange to deprive yourself of the benefit of the
patent license for this particular work, or (3) arrange, in a manner
consistent with the requirements of this License, to extend the patent
license to downstream recipients. "Knowingly relying" means you have
actual knowledge that, but for the patent license, your conveying the
covered work in a country, or your recipient's use of the covered work
in a country, would infringe one or more identifiable patents in that
country that you have reason to believe are valid.
If, pursuant to or in connection with a single transaction or
arrangement, you convey, or propagate by procuring conveyance of, a
covered work, and grant a patent license to some of the parties
receiving the covered work authorizing them to use, propagate, modify
or convey a specific copy of the covered work, then the patent license
you grant is automatically extended to all recipients of the covered
work and works based on it.
A patent license is "discriminatory" if it does not include within
the scope of its coverage, prohibits the exercise of, or is
conditioned on the non-exercise of one or more of the rights that are
specifically granted under this License. You may not convey a covered
work if you are a party to an arrangement with a third party that is
in the business of distributing software, under which you make payment
to the third party based on the extent of your activity of conveying
the work, and under which the third party grants, to any of the
parties who would receive the covered work from you, a discriminatory
patent license (a) in connection with copies of the covered work
conveyed by you (or copies made from those copies), or (b) primarily
for and in connection with specific products or compilations that
contain the covered work, unless you entered into that arrangement,
or that patent license was granted, prior to 28 March 2007.
Nothing in this License shall be construed as excluding or limiting
any implied license or other defenses to infringement that may
otherwise be available to you under applicable patent law.
12. No Surrender of Others' Freedom.
If conditions are imposed on you (whether by court order, agreement or
otherwise) that contradict the conditions of this License, they do not
excuse you from the conditions of this License. If you cannot convey a
covered work so as to satisfy simultaneously your obligations under this
License and any other pertinent obligations, then as a consequence you may
not convey it at all. For example, if you agree to terms that obligate you
to collect a royalty for further conveying from those to whom you convey
the Program, the only way you could satisfy both those terms and this
License would be to refrain entirely from conveying the Program.
13. Use with the GNU Affero General Public License.
Notwithstanding any other provision of this License, you have
permission to link or combine any covered work with a work licensed
under version 3 of the GNU Affero General Public License into a single
combined work, and to convey the resulting work. The terms of this
License will continue to apply to the part which is the covered work,
but the special requirements of the GNU Affero General Public License,
section 13, concerning interaction through a network will apply to the
combination as such.
14. Revised Versions of this License.
The Free Software Foundation may publish revised and/or new versions of
the GNU General Public License from time to time. Such new versions will
be similar in spirit to the present version, but may differ in detail to
address new problems or concerns.
Each version is given a distinguishing version number. If the
Program specifies that a certain numbered version of the GNU General
Public License "or any later version" applies to it, you have the
option of following the terms and conditions either of that numbered
version or of any later version published by the Free Software
Foundation. If the Program does not specify a version number of the
GNU General Public License, you may choose any version ever published
by the Free Software Foundation.
If the Program specifies that a proxy can decide which future
versions of the GNU General Public License can be used, that proxy's
public statement of acceptance of a version permanently authorizes you
to choose that version for the Program.
Later license versions may give you additional or different
permissions. However, no additional obligations are imposed on any
author or copyright holder as a result of your choosing to follow a
later version.
15. Disclaimer of Warranty.
THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
16. Limitation of Liability.
IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
SUCH DAMAGES.
17. Interpretation of Sections 15 and 16.
If the disclaimer of warranty and limitation of liability provided
above cannot be given local legal effect according to their terms,
reviewing courts shall apply local law that most closely approximates
an absolute waiver of all civil liability in connection with the
Program, unless a warranty or assumption of liability accompanies a
copy of the Program in return for a fee.
END OF TERMS AND CONDITIONS
How to Apply These Terms to Your New Programs
If you develop a new program, and you want it to be of the greatest
possible use to the public, the best way to achieve this is to make it
free software which everyone can redistribute and change under these terms.
To do so, attach the following notices to the program. It is safest
to attach them to the start of each source file to most effectively
state the exclusion of warranty; and each file should have at least
the "copyright" line and a pointer to where the full notice is found.
<one line to give the program's name and a brief idea of what it does.>
Copyright (C) <year> <name of author>
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
Also add information on how to contact you by electronic and paper mail.
If the program does terminal interaction, make it output a short
notice like this when it starts in an interactive mode:
<program> Copyright (C) <year> <name of author>
This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
This is free software, and you are welcome to redistribute it
under certain conditions; type `show c' for details.
The hypothetical commands `show w' and `show c' should show the appropriate
parts of the General Public License. Of course, your program's commands
might be different; for a GUI interface, you would use an "about box".
You should also get your employer (if you work as a programmer) or school,
if any, to sign a "copyright disclaimer" for the program, if necessary.
For more information on this, and how to apply and follow the GNU GPL, see
<http://www.gnu.org/licenses/>.
The GNU General Public License does not permit incorporating your program
into proprietary programs. If your program is a subroutine library, you
may consider it more useful to permit linking proprietary applications with
the library. If this is what you want to do, use the GNU Lesser General
Public License instead of this License. But first, please read
<http://www.gnu.org/philosophy/why-not-lgpl.html>.

164
src/Stockfish/README.md Normal file
View file

@ -0,0 +1,164 @@
<div align="center">
[![Stockfish][stockfish128-logo]][website-link]
<h3>Stockfish</h3>
A free and strong UCI chess engine.
<br>
<strong>[Explore Stockfish docs »][wiki-link]</strong>
<br>
<br>
[Report bug][issue-link]
·
[Open a discussion][discussions-link]
·
[Discord][discord-link]
·
[Blog][website-blog-link]
[![Build][build-badge]][build-link]
[![License][license-badge]][license-link]
<br>
[![Release][release-badge]][release-link]
[![Commits][commits-badge]][commits-link]
<br>
[![Website][website-badge]][website-link]
[![Fishtest][fishtest-badge]][fishtest-link]
[![Discord][discord-badge]][discord-link]
</div>
## Overview
[Stockfish][website-link] is a **free and strong UCI chess engine** derived from
Glaurung 2.1 that analyzes chess positions and computes the optimal moves.
Stockfish **does not include a graphical user interface** (GUI) that is required
to display a chessboard and to make it easy to input moves. These GUIs are
developed independently from Stockfish and are available online. **Read the
documentation for your GUI** of choice for information about how to use
Stockfish with it.
See also the Stockfish [documentation][wiki-usage-link] for further usage help.
## Files
This distribution of Stockfish consists of the following files:
* [README.md][readme-link], the file you are currently reading.
* [Copying.txt][license-link], a text file containing the GNU General Public
License version 3.
* [AUTHORS][authors-link], a text file with the list of authors for the project.
* [src][src-link], a subdirectory containing the full source code, including a
Makefile that can be used to compile Stockfish on Unix-like systems.
* a file with the .nnue extension, storing the neural network for the NNUE
evaluation. Binary distributions will have this file embedded.
## The UCI protocol
The [Universal Chess Interface][uci-link] (UCI) is a standard text-based protocol
used to communicate with a chess engine and is the recommended way to do so for
typical graphical user interfaces (GUI) or chess tools. Stockfish implements the
majority of its options.
Developers can see the default values for the UCI options available in Stockfish
by typing `./stockfish uci` in a terminal, but most users should typically use a
chess GUI to interact with Stockfish.
For more information on UCI or debug commands, see our [documentation][wiki-commands-link].
## Compiling Stockfish
Stockfish has support for 32 or 64-bit CPUs, certain hardware instructions,
big-endian machines such as Power PC, and other platforms.
On Unix-like systems, it should be easy to compile Stockfish directly from the
source code with the included Makefile in the folder `src`. In general, it is
recommended to run `make help` to see a list of make targets with corresponding
descriptions.
```
cd src
make -j build ARCH=x86-64-modern
```
Detailed compilation instructions for all platforms can be found in our
[documentation][wiki-compile-link].
## Contributing
### Donating hardware
Improving Stockfish requires a massive amount of testing. You can donate your
hardware resources by installing the [Fishtest Worker][worker-link] and viewing
the current tests on [Fishtest][fishtest-link].
### Improving the code
In the [chessprogramming wiki][programming-link], many techniques used in
Stockfish are explained with a lot of background information.
The [section on Stockfish][programmingsf-link] describes many features
and techniques used by Stockfish. However, it is generic rather than
focused on Stockfish's precise implementation.
The engine testing is done on [Fishtest][fishtest-link].
If you want to help improve Stockfish, please read this [guideline][guideline-link]
first, where the basics of Stockfish development are explained.
Discussions about Stockfish take place these days mainly in the Stockfish
[Discord server][discord-link]. This is also the best place to ask questions
about the codebase and how to improve it.
## Terms of use
Stockfish is free and distributed under the
[**GNU General Public License version 3**][license-link] (GPL v3). Essentially,
this means you are free to do almost exactly what you want with the program,
including distributing it among your friends, making it available for download
from your website, selling it (either by itself or as part of some bigger
software package), or using it as the starting point for a software project of
your own.
The only real limitation is that whenever you distribute Stockfish in some way,
you MUST always include the license and the full source code (or a pointer to
where the source code can be found) to generate the exact binary you are
distributing. If you make any changes to the source code, these changes must
also be made available under GPL v3.
[authors-link]: https://github.com/official-stockfish/Stockfish/blob/master/AUTHORS
[build-link]: https://github.com/official-stockfish/Stockfish/actions/workflows/stockfish.yml
[commits-link]: https://github.com/official-stockfish/Stockfish/commits/master
[discord-link]: https://discord.gg/GWDRS3kU6R
[issue-link]: https://github.com/official-stockfish/Stockfish/issues/new?assignees=&labels=&template=BUG-REPORT.yml
[discussions-link]: https://github.com/official-stockfish/Stockfish/discussions/new
[fishtest-link]: https://tests.stockfishchess.org/tests
[guideline-link]: https://github.com/glinscott/fishtest/wiki/Creating-my-first-test
[license-link]: https://github.com/official-stockfish/Stockfish/blob/master/Copying.txt
[programming-link]: https://www.chessprogramming.org/Main_Page
[programmingsf-link]: https://www.chessprogramming.org/Stockfish
[readme-link]: https://github.com/official-stockfish/Stockfish/blob/master/README.md
[release-link]: https://github.com/official-stockfish/Stockfish/releases/latest
[src-link]: https://github.com/official-stockfish/Stockfish/tree/master/src
[stockfish128-logo]: https://stockfishchess.org/images/logo/icon_128x128.png
[uci-link]: https://backscattering.de/chess/uci/
[website-link]: https://stockfishchess.org
[website-blog-link]: https://stockfishchess.org/blog/
[wiki-link]: https://github.com/official-stockfish/Stockfish/wiki
[wiki-usage-link]: https://github.com/official-stockfish/Stockfish/wiki/Download-and-usage
[wiki-compile-link]: https://github.com/official-stockfish/Stockfish/wiki/Compiling-from-source
[wiki-commands-link]: https://github.com/official-stockfish/Stockfish/wiki/Commands
[worker-link]: https://github.com/glinscott/fishtest/wiki/Running-the-worker
[build-badge]: https://img.shields.io/github/actions/workflow/status/official-stockfish/Stockfish/stockfish.yml?branch=master&style=for-the-badge&label=stockfish&logo=github
[commits-badge]: https://img.shields.io/github/commits-since/official-stockfish/Stockfish/latest?style=for-the-badge
[discord-badge]: https://img.shields.io/discord/435943710472011776?style=for-the-badge&label=discord&logo=Discord
[fishtest-badge]: https://img.shields.io/website?style=for-the-badge&down_color=red&down_message=Offline&label=Fishtest&up_color=success&up_message=Online&url=https%3A%2F%2Ftests.stockfishchess.org%2Ftests%2Ffinished
[license-badge]: https://img.shields.io/github/license/official-stockfish/Stockfish?style=for-the-badge&label=license&color=success
[release-badge]: https://img.shields.io/github/v/release/official-stockfish/Stockfish?style=for-the-badge&label=official%20release
[website-badge]: https://img.shields.io/website?style=for-the-badge&down_color=red&down_message=Offline&label=website&up_color=success&up_message=Online&url=https%3A%2F%2Fstockfishchess.org

267
src/Stockfish/Top CPU Contributors.txt Normal file
View file

@ -0,0 +1,267 @@
Contributors to Fishtest with >10,000 CPU hours, as of 2023-06-20.
Thank you!
Username CPU Hours Games played
------------------------------------------------------------------
noobpwnftw 37457426 2850540907
technologov 14135647 742892808
linrock 4423514 303254809
mlang 3026000 200065824
dew 1689162 100033738
okrout 1578136 148855886
pemo 1508508 48814305
grandphish2 1461406 91540343
TueRens 1194790 70400852
JojoM 947612 61773190
tvijlbrief 796125 51897690
sebastronomy 742434 38218524
mibere 703840 46867607
gvreuls 651026 42988582
oz 543438 39314736
cw 517858 34869755
fastgm 503862 30260818
leszek 467278 33514883
CSU_Dynasty 464940 31177118
ctoks 434416 28506889
crunchy 427035 27344275
maximmasiutin 424795 26577722
bcross 415722 29060963
olafm 395922 32268020
rpngn 348378 24560289
velislav 342567 22138992
Fisherman 327231 21829379
mgrabiak 300612 20608380
Dantist 296386 18031762
nordlandia 246201 16189678
robal 241300 15656382
marrco 234581 17714473
ncfish1 227517 15233777
glinscott 208125 13277240
drabel 204167 13930674
mhoram 202894 12601997
bking_US 198894 11876016
Thanar 179852 12365359
vdv 175544 9904472
spams 157128 10319326
sqrt2 147963 9724586
DesolatedDodo 146350 9536172
Calis007 143165 9478764
vdbergh 138650 9064413
CoffeeOne 137100 5024116
armo9494 136191 9460264
malala 136182 8002293
xoto 133759 9159372
davar 129023 8376525
DMBK 122960 8980062
dsmith 122059 7570238
amicic 119661 7938029
Data 113305 8220352
BrunoBanani 112960 7436849
CypressChess 108331 7759788
skiminki 107583 7218170
jcAEie 105675 8238962
MaZePallas 102823 6633619
sterni1971 100532 5880772
sunu 100167 7040199
zeryl 99331 6221261
thirdlife 99124 2242380
ElbertoOne 99028 7023771
cuistot 98853 6069816
bigpen0r 94809 6529203
brabos 92118 6186135
Wolfgang 91939 6105872
psk 89957 5984901
sschnee 88235 5268000
racerschmacer 85805 6122790
Fifis 85722 5709729
Dubslow 84986 6042456
Vizvezdenec 83761 5344740
0x3C33 82614 5271253
BRAVONE 81239 5054681
nssy 76497 5259388
jromang 76106 5236025
teddybaer 75125 5407666
tolkki963 74762 5149662
megaman7de 74351 4940352
Wencey 74181 4711488
Pking_cda 73776 5293873
yurikvelo 73150 5004382
markkulix 72607 5304642
Bobo1239 70579 4794999
solarlight 70517 5028306
dv8silencer 70287 3883992
manap 66273 4121774
tinker 64333 4268790
qurashee 61208 3429862
Mineta 59357 4418202
Spprtr 58723 3911011
AGI 58147 4325994
robnjr 57262 4053117
Freja 56938 3733019
MaxKlaxxMiner 56879 3423958
MarcusTullius 56746 3762951
ttruscott 56010 3680085
rkl 55132 4164467
renouve 53811 3501516
javran 53785 4627608
finfish 51360 3370515
eva42 51272 3599691
eastorwest 51117 3454811
rap 49985 3219146
pb00067 49733 3298934
OuaisBla 48626 3445134
ronaldjerum 47654 3240695
biffhero 46564 3111352
VoyagerOne 45476 3452465
jmdana 44893 3065205
maposora 44597 4039578
oryx 44570 3454238
speedycpu 43842 3003273
jbwiebe 43305 2805433
GPUex 42378 3133332
Antihistamine 41788 2761312
mhunt 41735 2691355
homyur 39893 2850481
gri 39871 2515779
Garf 37741 2999686
SC 37299 2731694
csnodgrass 36207 2688994
strelock 34716 2074055
szupaw 34102 2880346
EthanOConnor 33370 2090311
slakovv 32915 2021889
Gelma 31771 1551204
gopeto 31671 2060990
kdave 31157 2198362
manapbk 30987 1810399
Prcuvu 30377 2170122
anst 30301 2190091
jkiiski 30136 1904470
spcc 29925 1901692
hyperbolic.tom 29840 2017394
chuckstablers 29659 2093438
Pyafue 29650 1902349
belzedar94 28846 1811530
chriswk 26902 1868317
xwziegtm 26897 2124586
achambord 26582 1767323
Patrick_G 26276 1801617
yorkman 26193 1992080
Ulysses 25288 1689730
SFTUser 25182 1675689
nabildanial 24942 1519409
Sharaf_DG 24765 1786697
Maxim 24705 1502062
rodneyc 24376 1416402
agg177 23890 1395014
Goatminola 23763 1956036
Ente 23639 1671638
Jopo12321 23467 1483172
JanErik 23408 1703875
Isidor 23388 1680691
Norabor 23371 1603244
cisco2015 22920 1763301
jsys14 22824 1591906
Zirie 22542 1472937
team-oh 22272 1636708
Roady 22220 1465606
MazeOfGalious 21978 1629593
sg4032 21947 1643353
ianh2105 21725 1632562
xor12 21628 1680365
dex 21612 1467203
nesoneg 21494 1463031
user213718 21454 1404128
sphinx 21211 1384728
AndreasKrug 21097 1634811
jjoshua2 21001 1423089
Zake9298 20938 1565848
horst.prack 20878 1465656
0xB00B1ES 20590 1208666
j3corre 20405 941444
Adrian.Schmidt123 20316 1281436
wei 19973 1745989
notchris 19958 1800128
Serpensin 19840 1697528
Gaster319 19712 1677310
fishtester 19617 1257388
rstoesser 19569 1293588
eudhan 19274 1283717
votoanthuan 19108 1609992
vulcan 18871 1729392
Karpovbot 18766 1053178
qoo_charly_cai 18543 1284937
jundery 18445 1115855
ville 17883 1384026
chris 17698 1487385
purplefishies 17595 1092533
dju 17414 981289
iisiraider 17275 1049015
DragonLord 17014 1162790
redstone59 16842 1461780
Alb11747 16787 1213926
IgorLeMasson 16064 1147232
Karby 15982 979610
scuzzi 15757 968735
ako027ako 15671 1173203
Nikolay.IT 15154 1068349
Andrew Grant 15114 895539
Naven94 15054 834762
OssumOpossum 14857 1007129
ZacHFX 14783 1021842
enedene 14476 905279
bpfliegel 14233 882523
mpx86 14019 759568
jpulman 13982 870599
Skiff84 13826 721996
crocogoat 13803 1117422
Nesa92 13786 1114691
joster 13710 946160
mbeier 13650 1044928
Hjax 13535 915487
Nullvalue 13468 1140498
Dark_wizzie 13422 1007152
Rudolphous 13244 883140
pirt 13100 1009897
Machariel 13010 863104
infinigon 12991 943216
mabichito 12903 749391
thijsk 12886 722107
AdrianSA 12860 804972
Flopzee 12698 894821
korposzczur 12606 838168
fatmurphy 12547 853210
SapphireBrand 12416 969604
Oakwen 12399 844109
deflectooor 12386 579392
modolief 12386 896470
Farseer 12249 694108
Jackfish 12213 805008
pgontarz 12151 848794
dbernier 12103 860824
getraideBFF 12072 1024966
stocky 11954 699440
mschmidt 11941 803401
MooTheCow 11870 773598
FormazChar 11766 885707
whelanh 11557 245188
3cho 11494 1031076
infinity 11470 727027
aga 11412 695127
torbjo 11395 729145
Thomas A. Anderson 11372 732094
savage84 11358 670860
d64 11263 789184
ali-al-zhrani 11245 779246
snicolet 11106 869170
dapper 11032 771402
ols 10947 624903
Karmatron 10828 677458
basepi 10637 744851
Cubox 10621 826448
michaelrpg 10509 739239
OIVAS7572 10420 995586
jojo2357 10419 929708
WoodMan777 10380 873720
Garruk 10365 706465
dzjp 10343 732529

1043
src/Stockfish/src/Makefile Normal file
View file

File diff suppressed because it is too large Load diff

177
src/Stockfish/src/benchmark.cpp Normal file
View file

@ -0,0 +1,177 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "benchmark.h"
#include <fstream>
#include <iostream>
#include <istream>
#include <vector>
#include "position.h"
using namespace std;
namespace {
const vector<string> Defaults = {
"setoption name UCI_Chess960 value false",
"rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1",
"r3k2r/p1ppqpb1/bn2pnp1/3PN3/1p2P3/2N2Q1p/PPPBBPPP/R3K2R w KQkq - 0 10",
"8/2p5/3p4/KP5r/1R3p1k/8/4P1P1/8 w - - 0 11",
"4rrk1/pp1n3p/3q2pQ/2p1pb2/2PP4/2P3N1/P2B2PP/4RRK1 b - - 7 19",
"rq3rk1/ppp2ppp/1bnpb3/3N2B1/3NP3/7P/PPPQ1PP1/2KR3R w - - 7 14 moves d4e6",
"r1bq1r1k/1pp1n1pp/1p1p4/4p2Q/4Pp2/1BNP4/PPP2PPP/3R1RK1 w - - 2 14 moves g2g4",
"r3r1k1/2p2ppp/p1p1bn2/8/1q2P3/2NPQN2/PPP3PP/R4RK1 b - - 2 15",
"r1bbk1nr/pp3p1p/2n5/1N4p1/2Np1B2/8/PPP2PPP/2KR1B1R w kq - 0 13",
"r1bq1rk1/ppp1nppp/4n3/3p3Q/3P4/1BP1B3/PP1N2PP/R4RK1 w - - 1 16",
"4r1k1/r1q2ppp/ppp2n2/4P3/5Rb1/1N1BQ3/PPP3PP/R5K1 w - - 1 17",
"2rqkb1r/ppp2p2/2npb1p1/1N1Nn2p/2P1PP2/8/PP2B1PP/R1BQK2R b KQ - 0 11",
"r1bq1r1k/b1p1npp1/p2p3p/1p6/3PP3/1B2NN2/PP3PPP/R2Q1RK1 w - - 1 16",
"3r1rk1/p5pp/bpp1pp2/8/q1PP1P2/b3P3/P2NQRPP/1R2B1K1 b - - 6 22",
"r1q2rk1/2p1bppp/2Pp4/p6b/Q1PNp3/4B3/PP1R1PPP/2K4R w - - 2 18",
"4k2r/1pb2ppp/1p2p3/1R1p4/3P4/2r1PN2/P4PPP/1R4K1 b - - 3 22",
"3q2k1/pb3p1p/4pbp1/2r5/PpN2N2/1P2P2P/5PP1/Q2R2K1 b - - 4 26",
"6k1/6p1/6Pp/ppp5/3pn2P/1P3K2/1PP2P2/3N4 b - - 0 1",
"3b4/5kp1/1p1p1p1p/pP1PpP1P/P1P1P3/3KN3/8/8 w - - 0 1",
"2K5/p7/7P/5pR1/8/5k2/r7/8 w - - 0 1 moves g5g6 f3e3 g6g5 e3f3",
"8/6pk/1p6/8/PP3p1p/5P2/4KP1q/3Q4 w - - 0 1",
"7k/3p2pp/4q3/8/4Q3/5Kp1/P6b/8 w - - 0 1",
"8/2p5/8/2kPKp1p/2p4P/2P5/3P4/8 w - - 0 1",
"8/1p3pp1/7p/5P1P/2k3P1/8/2K2P2/8 w - - 0 1",
"8/pp2r1k1/2p1p3/3pP2p/1P1P1P1P/P5KR/8/8 w - - 0 1",
"8/3p4/p1bk3p/Pp6/1Kp1PpPp/2P2P1P/2P5/5B2 b - - 0 1",
"5k2/7R/4P2p/5K2/p1r2P1p/8/8/8 b - - 0 1",
"6k1/6p1/P6p/r1N5/5p2/7P/1b3PP1/4R1K1 w - - 0 1",
"1r3k2/4q3/2Pp3b/3Bp3/2Q2p2/1p1P2P1/1P2KP2/3N4 w - - 0 1",
"6k1/4pp1p/3p2p1/P1pPb3/R7/1r2P1PP/3B1P2/6K1 w - - 0 1",
"8/3p3B/5p2/5P2/p7/PP5b/k7/6K1 w - - 0 1",
"5rk1/q6p/2p3bR/1pPp1rP1/1P1Pp3/P3B1Q1/1K3P2/R7 w - - 93 90",
"4rrk1/1p1nq3/p7/2p1P1pp/3P2bp/3Q1Bn1/PPPB4/1K2R1NR w - - 40 21",
"r3k2r/3nnpbp/q2pp1p1/p7/Pp1PPPP1/4BNN1/1P5P/R2Q1RK1 w kq - 0 16",
"3Qb1k1/1r2ppb1/pN1n2q1/Pp1Pp1Pr/4P2p/4BP2/4B1R1/1R5K b - - 11 40",
"4k3/3q1r2/1N2r1b1/3ppN2/2nPP3/1B1R2n1/2R1Q3/3K4 w - - 5 1",
// 5-man positions
"8/8/8/8/5kp1/P7/8/1K1N4 w - - 0 1", // Kc2 - mate
"8/8/8/5N2/8/p7/8/2NK3k w - - 0 1", // Na2 - mate
"8/3k4/8/8/8/4B3/4KB2/2B5 w - - 0 1", // draw
// 6-man positions
"8/8/1P6/5pr1/8/4R3/7k/2K5 w - - 0 1", // Re5 - mate
"8/2p4P/8/kr6/6R1/8/8/1K6 w - - 0 1", // Ka2 - mate
"8/8/3P3k/8/1p6/8/1P6/1K3n2 b - - 0 1", // Nd2 - draw
// 7-man positions
"8/R7/2q5/8/6k1/8/1P5p/K6R w - - 0 124", // Draw
// Mate and stalemate positions
"6k1/3b3r/1p1p4/p1n2p2/1PPNpP1q/P3Q1p1/1R1RB1P1/5K2 b - - 0 1",
"r2r1n2/pp2bk2/2p1p2p/3q4/3PN1QP/2P3R1/P4PP1/5RK1 w - - 0 1",
"8/8/8/8/8/6k1/6p1/6K1 w - -",
"7k/7P/6K1/8/3B4/8/8/8 b - -",
// Chess 960
"setoption name UCI_Chess960 value true",
"bbqnnrkr/pppppppp/8/8/8/8/PPPPPPPP/BBQNNRKR w HFhf - 0 1 moves g2g3 d7d5 d2d4 c8h3 c1g5 e8d6 g5e7 f7f6",
"nqbnrkrb/pppppppp/8/8/8/8/PPPPPPPP/NQBNRKRB w KQkq - 0 1",
"setoption name UCI_Chess960 value false"
};
} // namespace
namespace Stockfish {
/// setup_bench() builds a list of UCI commands to be run by bench. There
/// are five parameters: TT size in MB, number of search threads that
/// should be used, the limit value spent for each position, a file name
/// where to look for positions in FEN format, the type of the limit:
/// depth, perft, nodes and movetime (in millisecs), and evaluation type
/// mixed (default), classical, NNUE.
///
/// bench -> search default positions up to depth 13
/// bench 64 1 15 -> search default positions up to depth 15 (TT = 64MB)
/// bench 64 4 5000 current movetime -> search current position with 4 threads for 5 sec
/// bench 64 1 100000 default nodes -> search default positions for 100K nodes each
/// bench 16 1 5 default perft -> run a perft 5 on default positions
vector<string> setup_bench(const Position& current, istream& is) {
vector<string> fens, list;
string go, token;
// Assign default values to missing arguments
string ttSize = (is >> token) ? token : "16";
string threads = (is >> token) ? token : "1";
string limit = (is >> token) ? token : "13";
string fenFile = (is >> token) ? token : "default";
string limitType = (is >> token) ? token : "depth";
string evalType = (is >> token) ? token : "mixed";
go = limitType == "eval" ? "eval" : "go " + limitType + " " + limit;
if (fenFile == "default")
fens = Defaults;
else if (fenFile == "current")
fens.push_back(current.fen());
else
{
string fen;
ifstream file(fenFile);
if (!file.is_open())
{
cerr << "Unable to open file " << fenFile << endl;
exit(EXIT_FAILURE);
}
while (getline(file, fen))
if (!fen.empty())
fens.push_back(fen);
file.close();
}
list.emplace_back("setoption name Threads value " + threads);
list.emplace_back("setoption name Hash value " + ttSize);
list.emplace_back("ucinewgame");
size_t posCounter = 0;
for (const string& fen : fens)
if (fen.find("setoption") != string::npos)
list.emplace_back(fen);
else
{
if (evalType == "classical" || (evalType == "mixed" && posCounter % 2 == 0))
list.emplace_back("setoption name Use NNUE value false");
else if (evalType == "NNUE" || (evalType == "mixed" && posCounter % 2 != 0))
list.emplace_back("setoption name Use NNUE value true");
list.emplace_back("position fen " + fen);
list.emplace_back(go);
++posCounter;
}
list.emplace_back("setoption name Use NNUE value true");
return list;
}
} // namespace Stockfish

34
src/Stockfish/src/benchmark.h Normal file
View file

@ -0,0 +1,34 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef BENCHMARK_H_INCLUDED
#define BENCHMARK_H_INCLUDED
#include <iosfwd>
#include <string>
#include <vector>
namespace Stockfish {
class Position;
std::vector<std::string> setup_bench(const Position&, std::istream&);
} // namespace Stockfish
#endif // #ifndef BENCHMARK_H_INCLUDED

172
src/Stockfish/src/bitbase.cpp Normal file
View file

@ -0,0 +1,172 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cassert>
#include <vector>
#include <bitset>
#include "bitboard.h"
#include "types.h"
namespace Stockfish {
namespace {
// There are 24 possible pawn squares: files A to D and ranks from 2 to 7.
// Positions with the pawn on files E to H will be mirrored before probing.
constexpr unsigned MAX_INDEX = 2*24*64*64; // stm * psq * wksq * bksq = 196608
std::bitset<MAX_INDEX> KPKBitbase;
// A KPK bitbase index is an integer in [0, IndexMax] range
//
// Information is mapped in a way that minimizes the number of iterations:
//
// bit 0- 5: white king square (from SQ_A1 to SQ_H8)
// bit 6-11: black king square (from SQ_A1 to SQ_H8)
// bit 12: side to move (WHITE or BLACK)
// bit 13-14: white pawn file (from FILE_A to FILE_D)
// bit 15-17: white pawn RANK_7 - rank (from RANK_7 - RANK_7 to RANK_7 - RANK_2)
unsigned index(Color stm, Square bksq, Square wksq, Square psq) {
return int(wksq) | (bksq << 6) | (stm << 12) | (file_of(psq) << 13) | ((RANK_7 - rank_of(psq)) << 15);
}
enum Result {
INVALID = 0,
UNKNOWN = 1,
DRAW = 2,
WIN = 4
};
Result& operator|=(Result& r, Result v) { return r = Result(r | v); }
struct KPKPosition {
KPKPosition() = default;
explicit KPKPosition(unsigned idx);
operator Result() const { return result; }
Result classify(const std::vector<KPKPosition>& db);
Color stm;
Square ksq[COLOR_NB], psq;
Result result;
};
} // namespace
bool Bitbases::probe(Square wksq, Square wpsq, Square bksq, Color stm) {
assert(file_of(wpsq) <= FILE_D);
return KPKBitbase[index(stm, bksq, wksq, wpsq)];
}
void Bitbases::init() {
std::vector<KPKPosition> db(MAX_INDEX);
unsigned idx, repeat = 1;
// Initialize db with known win / draw positions
for (idx = 0; idx < MAX_INDEX; ++idx)
db[idx] = KPKPosition(idx);
// Iterate through the positions until none of the unknown positions can be
// changed to either wins or draws (15 cycles needed).
while (repeat)
for (repeat = idx = 0; idx < MAX_INDEX; ++idx)
repeat |= (db[idx] == UNKNOWN && db[idx].classify(db) != UNKNOWN);
// Fill the bitbase with the decisive results
for (idx = 0; idx < MAX_INDEX; ++idx)
if (db[idx] == WIN)
KPKBitbase.set(idx);
}
namespace {
KPKPosition::KPKPosition(unsigned idx) {
ksq[WHITE] = Square((idx >> 0) & 0x3F);
ksq[BLACK] = Square((idx >> 6) & 0x3F);
stm = Color ((idx >> 12) & 0x01);
psq = make_square(File((idx >> 13) & 0x3), Rank(RANK_7 - ((idx >> 15) & 0x7)));
// Invalid if two pieces are on the same square or if a king can be captured
if ( distance(ksq[WHITE], ksq[BLACK]) <= 1
|| ksq[WHITE] == psq
|| ksq[BLACK] == psq
|| (stm == WHITE && (pawn_attacks_bb(WHITE, psq) & ksq[BLACK])))
result = INVALID;
// Win if the pawn can be promoted without getting captured
else if ( stm == WHITE
&& rank_of(psq) == RANK_7
&& ksq[WHITE] != psq + NORTH
&& ( distance(ksq[BLACK], psq + NORTH) > 1
|| (distance(ksq[WHITE], psq + NORTH) == 1)))
result = WIN;
// Draw if it is stalemate or the black king can capture the pawn
else if ( stm == BLACK
&& ( !(attacks_bb<KING>(ksq[BLACK]) & ~(attacks_bb<KING>(ksq[WHITE]) | pawn_attacks_bb(WHITE, psq)))
|| (attacks_bb<KING>(ksq[BLACK]) & ~attacks_bb<KING>(ksq[WHITE]) & psq)))
result = DRAW;
// Position will be classified later
else
result = UNKNOWN;
}
Result KPKPosition::classify(const std::vector<KPKPosition>& db) {
// White to move: If one move leads to a position classified as WIN, the result
// of the current position is WIN. If all moves lead to positions classified
// as DRAW, the current position is classified as DRAW, otherwise the current
// position is classified as UNKNOWN.
//
// Black to move: If one move leads to a position classified as DRAW, the result
// of the current position is DRAW. If all moves lead to positions classified
// as WIN, the position is classified as WIN, otherwise the current position is
// classified as UNKNOWN.
const Result Good = (stm == WHITE ? WIN : DRAW);
const Result Bad = (stm == WHITE ? DRAW : WIN);
Result r = INVALID;
Bitboard b = attacks_bb<KING>(ksq[stm]);
while (b)
r |= stm == WHITE ? db[index(BLACK, ksq[BLACK], pop_lsb(b), psq)]
: db[index(WHITE, pop_lsb(b), ksq[WHITE], psq)];
if (stm == WHITE)
{
if (rank_of(psq) < RANK_7) // Single push
r |= db[index(BLACK, ksq[BLACK], ksq[WHITE], psq + NORTH)];
if ( rank_of(psq) == RANK_2 // Double push
&& psq + NORTH != ksq[WHITE]
&& psq + NORTH != ksq[BLACK])
r |= db[index(BLACK, ksq[BLACK], ksq[WHITE], psq + NORTH + NORTH)];
}
return result = r & Good ? Good : r & UNKNOWN ? UNKNOWN : Bad;
}
} // namespace
} // namespace Stockfish

218
src/Stockfish/src/bitboard.cpp Normal file
View file

@ -0,0 +1,218 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm>
#include <bitset>
#include "bitboard.h"
#include "misc.h"
namespace Stockfish {
uint8_t PopCnt16[1 << 16];
uint8_t SquareDistance[SQUARE_NB][SQUARE_NB];
Bitboard LineBB[SQUARE_NB][SQUARE_NB];
Bitboard BetweenBB[SQUARE_NB][SQUARE_NB];
Bitboard PseudoAttacks[PIECE_TYPE_NB][SQUARE_NB];
Bitboard PawnAttacks[COLOR_NB][SQUARE_NB];
Magic RookMagics[SQUARE_NB];
Magic BishopMagics[SQUARE_NB];
namespace {
Bitboard RookTable[0x19000]; // To store rook attacks
Bitboard BishopTable[0x1480]; // To store bishop attacks
void init_magics(PieceType pt, Bitboard table[], Magic magics[]);
}
/// safe_destination() returns the bitboard of target square for the given step
/// from the given square. If the step is off the board, returns empty bitboard.
inline Bitboard safe_destination(Square s, int step) {
Square to = Square(s + step);
return is_ok(to) && distance(s, to) <= 2 ? square_bb(to) : Bitboard(0);
}
/// Bitboards::pretty() returns an ASCII representation of a bitboard suitable
/// to be printed to standard output. Useful for debugging.
std::string Bitboards::pretty(Bitboard b) {
std::string s = "+---+---+---+---+---+---+---+---+\n";
for (Rank r = RANK_8; r >= RANK_1; --r)
{
for (File f = FILE_A; f <= FILE_H; ++f)
s += b & make_square(f, r) ? "| X " : "| ";
s += "| " + std::to_string(1 + r) + "\n+---+---+---+---+---+---+---+---+\n";
}
s += " a b c d e f g h\n";
return s;
}
/// Bitboards::init() initializes various bitboard tables. It is called at
/// startup and relies on global objects to be already zero-initialized.
void Bitboards::init() {
for (unsigned i = 0; i < (1 << 16); ++i)
PopCnt16[i] = uint8_t(std::bitset<16>(i).count());
for (Square s1 = SQ_A1; s1 <= SQ_H8; ++s1)
for (Square s2 = SQ_A1; s2 <= SQ_H8; ++s2)
SquareDistance[s1][s2] = std::max(distance<File>(s1, s2), distance<Rank>(s1, s2));
init_magics(ROOK, RookTable, RookMagics);
init_magics(BISHOP, BishopTable, BishopMagics);
for (Square s1 = SQ_A1; s1 <= SQ_H8; ++s1)
{
PawnAttacks[WHITE][s1] = pawn_attacks_bb<WHITE>(square_bb(s1));
PawnAttacks[BLACK][s1] = pawn_attacks_bb<BLACK>(square_bb(s1));
for (int step : {-9, -8, -7, -1, 1, 7, 8, 9} )
PseudoAttacks[KING][s1] |= safe_destination(s1, step);
for (int step : {-17, -15, -10, -6, 6, 10, 15, 17} )
PseudoAttacks[KNIGHT][s1] |= safe_destination(s1, step);
PseudoAttacks[QUEEN][s1] = PseudoAttacks[BISHOP][s1] = attacks_bb<BISHOP>(s1, 0);
PseudoAttacks[QUEEN][s1] |= PseudoAttacks[ ROOK][s1] = attacks_bb< ROOK>(s1, 0);
for (PieceType pt : { BISHOP, ROOK })
for (Square s2 = SQ_A1; s2 <= SQ_H8; ++s2)
{
if (PseudoAttacks[pt][s1] & s2)
{
LineBB[s1][s2] = (attacks_bb(pt, s1, 0) & attacks_bb(pt, s2, 0)) | s1 | s2;
BetweenBB[s1][s2] = (attacks_bb(pt, s1, square_bb(s2)) & attacks_bb(pt, s2, square_bb(s1)));
}
BetweenBB[s1][s2] |= s2;
}
}
}
namespace {
Bitboard sliding_attack(PieceType pt, Square sq, Bitboard occupied) {
Bitboard attacks = 0;
Direction RookDirections[4] = {NORTH, SOUTH, EAST, WEST};
Direction BishopDirections[4] = {NORTH_EAST, SOUTH_EAST, SOUTH_WEST, NORTH_WEST};
for (Direction d : (pt == ROOK ? RookDirections : BishopDirections))
{
Square s = sq;
while (safe_destination(s, d) && !(occupied & s))
attacks |= (s += d);
}
return attacks;
}
// init_magics() computes all rook and bishop attacks at startup. Magic
// bitboards are used to look up attacks of sliding pieces. As a reference see
// www.chessprogramming.org/Magic_Bitboards. In particular, here we use the so
// called "fancy" approach.
void init_magics(PieceType pt, Bitboard table[], Magic magics[]) {
// Optimal PRNG seeds to pick the correct magics in the shortest time
int seeds[][RANK_NB] = { { 8977, 44560, 54343, 38998, 5731, 95205, 104912, 17020 },
{ 728, 10316, 55013, 32803, 12281, 15100, 16645, 255 } };
Bitboard occupancy[4096], reference[4096], edges, b;
int epoch[4096] = {}, cnt = 0, size = 0;
for (Square s = SQ_A1; s <= SQ_H8; ++s)
{
// Board edges are not considered in the relevant occupancies
edges = ((Rank1BB | Rank8BB) & ~rank_bb(s)) | ((FileABB | FileHBB) & ~file_bb(s));
// Given a square 's', the mask is the bitboard of sliding attacks from
// 's' computed on an empty board. The index must be big enough to contain
// all the attacks for each possible subset of the mask and so is 2 power
// the number of 1s of the mask. Hence we deduce the size of the shift to
// apply to the 64 or 32 bits word to get the index.
Magic& m = magics[s];
m.mask = sliding_attack(pt, s, 0) & ~edges;
m.shift = (Is64Bit ? 64 : 32) - popcount(m.mask);
// Set the offset for the attacks table of the square. We have individual
// table sizes for each square with "Fancy Magic Bitboards".
m.attacks = s == SQ_A1 ? table : magics[s - 1].attacks + size;
// Use Carry-Rippler trick to enumerate all subsets of masks[s] and
// store the corresponding sliding attack bitboard in reference[].
b = size = 0;
do {
occupancy[size] = b;
reference[size] = sliding_attack(pt, s, b);
if (HasPext)
m.attacks[pext(b, m.mask)] = reference[size];
size++;
b = (b - m.mask) & m.mask;
} while (b);
if (HasPext)
continue;
PRNG rng(seeds[Is64Bit][rank_of(s)]);
// Find a magic for square 's' picking up an (almost) random number
// until we find the one that passes the verification test.
for (int i = 0; i < size; )
{
for (m.magic = 0; popcount((m.magic * m.mask) >> 56) < 6; )
m.magic = rng.sparse_rand<Bitboard>();
// A good magic must map every possible occupancy to an index that
// looks up the correct sliding attack in the attacks[s] database.
// Note that we build up the database for square 's' as a side
// effect of verifying the magic. Keep track of the attempt count
// and save it in epoch[], little speed-up trick to avoid resetting
// m.attacks[] after every failed attempt.
for (++cnt, i = 0; i < size; ++i)
{
unsigned idx = m.index(occupancy[i]);
if (epoch[idx] < cnt)
{
epoch[idx] = cnt;
m.attacks[idx] = reference[i];
}
else if (m.attacks[idx] != reference[i])
break;
}
}
}
}
}
} // namespace Stockfish

450
src/Stockfish/src/bitboard.h Normal file
View file

@ -0,0 +1,450 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef BITBOARD_H_INCLUDED
#define BITBOARD_H_INCLUDED
#include <string>
#include "types.h"
namespace Stockfish {
namespace Bitbases {
void init();
bool probe(Square wksq, Square wpsq, Square bksq, Color us);
} // namespace Stockfish::Bitbases
namespace Bitboards {
void init();
std::string pretty(Bitboard b);
} // namespace Stockfish::Bitboards
constexpr Bitboard AllSquares = ~Bitboard(0);
constexpr Bitboard DarkSquares = 0xAA55AA55AA55AA55ULL;
constexpr Bitboard FileABB = 0x0101010101010101ULL;
constexpr Bitboard FileBBB = FileABB << 1;
constexpr Bitboard FileCBB = FileABB << 2;
constexpr Bitboard FileDBB = FileABB << 3;
constexpr Bitboard FileEBB = FileABB << 4;
constexpr Bitboard FileFBB = FileABB << 5;
constexpr Bitboard FileGBB = FileABB << 6;
constexpr Bitboard FileHBB = FileABB << 7;
constexpr Bitboard Rank1BB = 0xFF;
constexpr Bitboard Rank2BB = Rank1BB << (8 * 1);
constexpr Bitboard Rank3BB = Rank1BB << (8 * 2);
constexpr Bitboard Rank4BB = Rank1BB << (8 * 3);
constexpr Bitboard Rank5BB = Rank1BB << (8 * 4);
constexpr Bitboard Rank6BB = Rank1BB << (8 * 5);
constexpr Bitboard Rank7BB = Rank1BB << (8 * 6);
constexpr Bitboard Rank8BB = Rank1BB << (8 * 7);
constexpr Bitboard QueenSide = FileABB | FileBBB | FileCBB | FileDBB;
constexpr Bitboard CenterFiles = FileCBB | FileDBB | FileEBB | FileFBB;
constexpr Bitboard KingSide = FileEBB | FileFBB | FileGBB | FileHBB;
constexpr Bitboard Center = (FileDBB | FileEBB) & (Rank4BB | Rank5BB);
constexpr Bitboard KingFlank[FILE_NB] = {
QueenSide ^ FileDBB, QueenSide, QueenSide,
CenterFiles, CenterFiles,
KingSide, KingSide, KingSide ^ FileEBB
};
extern uint8_t PopCnt16[1 << 16];
extern uint8_t SquareDistance[SQUARE_NB][SQUARE_NB];
extern Bitboard BetweenBB[SQUARE_NB][SQUARE_NB];
extern Bitboard LineBB[SQUARE_NB][SQUARE_NB];
extern Bitboard PseudoAttacks[PIECE_TYPE_NB][SQUARE_NB];
extern Bitboard PawnAttacks[COLOR_NB][SQUARE_NB];
/// Magic holds all magic bitboards relevant data for a single square
struct Magic {
Bitboard mask;
Bitboard magic;
Bitboard* attacks;
unsigned shift;
// Compute the attack's index using the 'magic bitboards' approach
unsigned index(Bitboard occupied) const {
if (HasPext)
return unsigned(pext(occupied, mask));
if (Is64Bit)
return unsigned(((occupied & mask) * magic) >> shift);
unsigned lo = unsigned(occupied) & unsigned(mask);
unsigned hi = unsigned(occupied >> 32) & unsigned(mask >> 32);
return (lo * unsigned(magic) ^ hi * unsigned(magic >> 32)) >> shift;
}
};
extern Magic RookMagics[SQUARE_NB];
extern Magic BishopMagics[SQUARE_NB];
inline Bitboard square_bb(Square s) {
assert(is_ok(s));
return (1ULL << s);
}
/// Overloads of bitwise operators between a Bitboard and a Square for testing
/// whether a given bit is set in a bitboard, and for setting and clearing bits.
inline Bitboard operator&( Bitboard b, Square s) { return b & square_bb(s); }
inline Bitboard operator|( Bitboard b, Square s) { return b | square_bb(s); }
inline Bitboard operator^( Bitboard b, Square s) { return b ^ square_bb(s); }
inline Bitboard& operator|=(Bitboard& b, Square s) { return b |= square_bb(s); }
inline Bitboard& operator^=(Bitboard& b, Square s) { return b ^= square_bb(s); }
inline Bitboard operator&(Square s, Bitboard b) { return b & s; }
inline Bitboard operator|(Square s, Bitboard b) { return b | s; }
inline Bitboard operator^(Square s, Bitboard b) { return b ^ s; }
inline Bitboard operator|(Square s1, Square s2) { return square_bb(s1) | s2; }
constexpr bool more_than_one(Bitboard b) {
return b & (b - 1);
}
constexpr bool opposite_colors(Square s1, Square s2) {
return (s1 + rank_of(s1) + s2 + rank_of(s2)) & 1;
}
/// rank_bb() and file_bb() return a bitboard representing all the squares on
/// the given file or rank.
constexpr Bitboard rank_bb(Rank r) {
return Rank1BB << (8 * r);
}
constexpr Bitboard rank_bb(Square s) {
return rank_bb(rank_of(s));
}
constexpr Bitboard file_bb(File f) {
return FileABB << f;
}
constexpr Bitboard file_bb(Square s) {
return file_bb(file_of(s));
}
/// shift() moves a bitboard one or two steps as specified by the direction D
template<Direction D>
constexpr Bitboard shift(Bitboard b) {
return D == NORTH ? b << 8 : D == SOUTH ? b >> 8
: D == NORTH+NORTH? b <<16 : D == SOUTH+SOUTH? b >>16
: D == EAST ? (b & ~FileHBB) << 1 : D == WEST ? (b & ~FileABB) >> 1
: D == NORTH_EAST ? (b & ~FileHBB) << 9 : D == NORTH_WEST ? (b & ~FileABB) << 7
: D == SOUTH_EAST ? (b & ~FileHBB) >> 7 : D == SOUTH_WEST ? (b & ~FileABB) >> 9
: 0;
}
/// pawn_attacks_bb() returns the squares attacked by pawns of the given color
/// from the squares in the given bitboard.
template<Color C>
constexpr Bitboard pawn_attacks_bb(Bitboard b) {
return C == WHITE ? shift<NORTH_WEST>(b) | shift<NORTH_EAST>(b)
: shift<SOUTH_WEST>(b) | shift<SOUTH_EAST>(b);
}
inline Bitboard pawn_attacks_bb(Color c, Square s) {
assert(is_ok(s));
return PawnAttacks[c][s];
}
/// pawn_double_attacks_bb() returns the squares doubly attacked by pawns of the
/// given color from the squares in the given bitboard.
template<Color C>
constexpr Bitboard pawn_double_attacks_bb(Bitboard b) {
return C == WHITE ? shift<NORTH_WEST>(b) & shift<NORTH_EAST>(b)
: shift<SOUTH_WEST>(b) & shift<SOUTH_EAST>(b);
}
/// adjacent_files_bb() returns a bitboard representing all the squares on the
/// adjacent files of a given square.
constexpr Bitboard adjacent_files_bb(Square s) {
return shift<EAST>(file_bb(s)) | shift<WEST>(file_bb(s));
}
/// line_bb() returns a bitboard representing an entire line (from board edge
/// to board edge) that intersects the two given squares. If the given squares
/// are not on a same file/rank/diagonal, the function returns 0. For instance,
/// line_bb(SQ_C4, SQ_F7) will return a bitboard with the A2-G8 diagonal.
inline Bitboard line_bb(Square s1, Square s2) {
assert(is_ok(s1) && is_ok(s2));
return LineBB[s1][s2];
}
/// between_bb(s1, s2) returns a bitboard representing the squares in the semi-open
/// segment between the squares s1 and s2 (excluding s1 but including s2). If the
/// given squares are not on a same file/rank/diagonal, it returns s2. For instance,
/// between_bb(SQ_C4, SQ_F7) will return a bitboard with squares D5, E6 and F7, but
/// between_bb(SQ_E6, SQ_F8) will return a bitboard with the square F8. This trick
/// allows to generate non-king evasion moves faster: the defending piece must either
/// interpose itself to cover the check or capture the checking piece.
inline Bitboard between_bb(Square s1, Square s2) {
assert(is_ok(s1) && is_ok(s2));
return BetweenBB[s1][s2];
}
/// forward_ranks_bb() returns a bitboard representing the squares on the ranks in
/// front of the given one, from the point of view of the given color. For instance,
/// forward_ranks_bb(BLACK, SQ_D3) will return the 16 squares on ranks 1 and 2.
constexpr Bitboard forward_ranks_bb(Color c, Square s) {
return c == WHITE ? ~Rank1BB << 8 * relative_rank(WHITE, s)
: ~Rank8BB >> 8 * relative_rank(BLACK, s);
}
/// forward_file_bb() returns a bitboard representing all the squares along the
/// line in front of the given one, from the point of view of the given color.
constexpr Bitboard forward_file_bb(Color c, Square s) {
return forward_ranks_bb(c, s) & file_bb(s);
}
/// pawn_attack_span() returns a bitboard representing all the squares that can
/// be attacked by a pawn of the given color when it moves along its file, starting
/// from the given square.
constexpr Bitboard pawn_attack_span(Color c, Square s) {
return forward_ranks_bb(c, s) & adjacent_files_bb(s);
}
/// passed_pawn_span() returns a bitboard which can be used to test if a pawn of
/// the given color and on the given square is a passed pawn.
constexpr Bitboard passed_pawn_span(Color c, Square s) {
return pawn_attack_span(c, s) | forward_file_bb(c, s);
}
/// aligned() returns true if the squares s1, s2 and s3 are aligned either on a
/// straight or on a diagonal line.
inline bool aligned(Square s1, Square s2, Square s3) {
return line_bb(s1, s2) & s3;
}
/// distance() functions return the distance between x and y, defined as the
/// number of steps for a king in x to reach y.
template<typename T1 = Square> inline int distance(Square x, Square y);
template<> inline int distance<File>(Square x, Square y) { return std::abs(file_of(x) - file_of(y)); }
template<> inline int distance<Rank>(Square x, Square y) { return std::abs(rank_of(x) - rank_of(y)); }
template<> inline int distance<Square>(Square x, Square y) { return SquareDistance[x][y]; }
inline int edge_distance(File f) { return std::min(f, File(FILE_H - f)); }
inline int edge_distance(Rank r) { return std::min(r, Rank(RANK_8 - r)); }
/// attacks_bb(Square) returns the pseudo attacks of the give piece type
/// assuming an empty board.
template<PieceType Pt>
inline Bitboard attacks_bb(Square s) {
assert((Pt != PAWN) && (is_ok(s)));
return PseudoAttacks[Pt][s];
}
/// attacks_bb(Square, Bitboard) returns the attacks by the given piece
/// assuming the board is occupied according to the passed Bitboard.
/// Sliding piece attacks do not continue passed an occupied square.
template<PieceType Pt>
inline Bitboard attacks_bb(Square s, Bitboard occupied) {
assert((Pt != PAWN) && (is_ok(s)));
switch (Pt)
{
case BISHOP: return BishopMagics[s].attacks[BishopMagics[s].index(occupied)];
case ROOK : return RookMagics[s].attacks[ RookMagics[s].index(occupied)];
case QUEEN : return attacks_bb<BISHOP>(s, occupied) | attacks_bb<ROOK>(s, occupied);
default : return PseudoAttacks[Pt][s];
}
}
inline Bitboard attacks_bb(PieceType pt, Square s, Bitboard occupied) {
assert((pt != PAWN) && (is_ok(s)));
switch (pt)
{
case BISHOP: return attacks_bb<BISHOP>(s, occupied);
case ROOK : return attacks_bb< ROOK>(s, occupied);
case QUEEN : return attacks_bb<BISHOP>(s, occupied) | attacks_bb<ROOK>(s, occupied);
default : return PseudoAttacks[pt][s];
}
}
/// popcount() counts the number of non-zero bits in a bitboard
inline int popcount(Bitboard b) {
#ifndef USE_POPCNT
union { Bitboard bb; uint16_t u[4]; } v = { b };
return PopCnt16[v.u[0]] + PopCnt16[v.u[1]] + PopCnt16[v.u[2]] + PopCnt16[v.u[3]];
#elif defined(_MSC_VER) || defined(__INTEL_COMPILER)
return (int)_mm_popcnt_u64(b);
#else // Assumed gcc or compatible compiler
return __builtin_popcountll(b);
#endif
}
/// lsb() and msb() return the least/most significant bit in a non-zero bitboard
#if defined(__GNUC__) // GCC, Clang, ICC
inline Square lsb(Bitboard b) {
assert(b);
return Square(__builtin_ctzll(b));
}
inline Square msb(Bitboard b) {
assert(b);
return Square(63 ^ __builtin_clzll(b));
}
#elif defined(_MSC_VER) // MSVC
#ifdef _WIN64 // MSVC, WIN64
inline Square lsb(Bitboard b) {
assert(b);
unsigned long idx;
_BitScanForward64(&idx, b);
return (Square) idx;
}
inline Square msb(Bitboard b) {
assert(b);
unsigned long idx;
_BitScanReverse64(&idx, b);
return (Square) idx;
}
#else // MSVC, WIN32
inline Square lsb(Bitboard b) {
assert(b);
unsigned long idx;
if (b & 0xffffffff) {
_BitScanForward(&idx, int32_t(b));
return Square(idx);
} else {
_BitScanForward(&idx, int32_t(b >> 32));
return Square(idx + 32);
}
}
inline Square msb(Bitboard b) {
assert(b);
unsigned long idx;
if (b >> 32) {
_BitScanReverse(&idx, int32_t(b >> 32));
return Square(idx + 32);
} else {
_BitScanReverse(&idx, int32_t(b));
return Square(idx);
}
}
#endif
#else // Compiler is neither GCC nor MSVC compatible
#error "Compiler not supported."
#endif
/// least_significant_square_bb() returns the bitboard of the least significant
/// square of a non-zero bitboard. It is equivalent to square_bb(lsb(bb)).
inline Bitboard least_significant_square_bb(Bitboard b) {
assert(b);
return b & -b;
}
/// pop_lsb() finds and clears the least significant bit in a non-zero bitboard
inline Square pop_lsb(Bitboard& b) {
assert(b);
const Square s = lsb(b);
b &= b - 1;
return s;
}
/// frontmost_sq() returns the most advanced square for the given color,
/// requires a non-zero bitboard.
inline Square frontmost_sq(Color c, Bitboard b) {
assert(b);
return c == WHITE ? msb(b) : lsb(b);
}
} // namespace Stockfish
#endif // #ifndef BITBOARD_H_INCLUDED

747
src/Stockfish/src/endgame.cpp Normal file
View file

@ -0,0 +1,747 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cassert>
#include "bitboard.h"
#include "endgame.h"
#include "movegen.h"
namespace Stockfish {
namespace {
// Used to drive the king towards the edge of the board
// in KX vs K and KQ vs KR endgames.
// Values range from 27 (center squares) to 90 (in the corners)
inline int push_to_edge(Square s) {
int rd = edge_distance(rank_of(s)), fd = edge_distance(file_of(s));
return 90 - (7 * fd * fd / 2 + 7 * rd * rd / 2);
}
// Used to drive the king towards A1H8 corners in KBN vs K endgames.
// Values range from 0 on A8H1 diagonal to 7 in A1H8 corners
inline int push_to_corner(Square s) {
return abs(7 - rank_of(s) - file_of(s));
}
// Drive a piece close to or away from another piece
inline int push_close(Square s1, Square s2) { return 140 - 20 * distance(s1, s2); }
inline int push_away(Square s1, Square s2) { return 120 - push_close(s1, s2); }
#ifndef NDEBUG
bool verify_material(const Position& pos, Color c, Value npm, int pawnsCnt) {
return pos.non_pawn_material(c) == npm && pos.count<PAWN>(c) == pawnsCnt;
}
#endif
// Map the square as if strongSide is white and strongSide's only pawn
// is on the left half of the board.
Square normalize(const Position& pos, Color strongSide, Square sq) {
assert(pos.count<PAWN>(strongSide) == 1);
if (file_of(pos.square<PAWN>(strongSide)) >= FILE_E)
sq = flip_file(sq);
return strongSide == WHITE ? sq : flip_rank(sq);
}
} // namespace
namespace Endgames {
std::pair<Map<Value>, Map<ScaleFactor>> maps;
void init() {
add<KPK>("KPK");
add<KNNK>("KNNK");
add<KBNK>("KBNK");
add<KRKP>("KRKP");
add<KRKB>("KRKB");
add<KRKN>("KRKN");
add<KQKP>("KQKP");
add<KQKR>("KQKR");
add<KNNKP>("KNNKP");
add<KRPKR>("KRPKR");
add<KRPKB>("KRPKB");
add<KBPKB>("KBPKB");
add<KBPKN>("KBPKN");
add<KBPPKB>("KBPPKB");
add<KRPPKRP>("KRPPKRP");
}
}
/// Mate with KX vs K. This function is used to evaluate positions with
/// king and plenty of material vs a lone king. It simply gives the
/// attacking side a bonus for driving the defending king towards the edge
/// of the board, and for keeping the distance between the two kings small.
template<>
Value Endgame<KXK>::operator()(const Position& pos) const {
assert(verify_material(pos, weakSide, VALUE_ZERO, 0));
assert(!pos.checkers()); // Eval is never called when in check
// Stalemate detection with lone king
if (pos.side_to_move() == weakSide && !MoveList<LEGAL>(pos).size())
return VALUE_DRAW;
Square strongKing = pos.square<KING>(strongSide);
Square weakKing = pos.square<KING>(weakSide);
Value result = pos.non_pawn_material(strongSide)
+ pos.count<PAWN>(strongSide) * PawnValueEg
+ push_to_edge(weakKing)
+ push_close(strongKing, weakKing);
if ( pos.count<QUEEN>(strongSide)
|| pos.count<ROOK>(strongSide)
||(pos.count<BISHOP>(strongSide) && pos.count<KNIGHT>(strongSide))
|| ( (pos.pieces(strongSide, BISHOP) & ~DarkSquares)
&& (pos.pieces(strongSide, BISHOP) & DarkSquares)))
result = std::min(result + VALUE_KNOWN_WIN, VALUE_TB_WIN_IN_MAX_PLY - 1);
return strongSide == pos.side_to_move() ? result : -result;
}
/// Mate with KBN vs K. This is similar to KX vs K, but we have to drive the
/// defending king towards a corner square that our bishop attacks.
template<>
Value Endgame<KBNK>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, KnightValueMg + BishopValueMg, 0));
assert(verify_material(pos, weakSide, VALUE_ZERO, 0));
Square strongKing = pos.square<KING>(strongSide);
Square strongBishop = pos.square<BISHOP>(strongSide);
Square weakKing = pos.square<KING>(weakSide);
// If our bishop does not attack A1/H8, we flip the enemy king square
// to drive to opposite corners (A8/H1).
Value result = (VALUE_KNOWN_WIN + 3520)
+ push_close(strongKing, weakKing)
+ 420 * push_to_corner(opposite_colors(strongBishop, SQ_A1) ? flip_file(weakKing) : weakKing);
assert(abs(result) < VALUE_TB_WIN_IN_MAX_PLY);
return strongSide == pos.side_to_move() ? result : -result;
}
/// KP vs K. This endgame is evaluated with the help of a bitbase
template<>
Value Endgame<KPK>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, VALUE_ZERO, 1));
assert(verify_material(pos, weakSide, VALUE_ZERO, 0));
// Assume strongSide is white and the pawn is on files A-D
Square strongKing = normalize(pos, strongSide, pos.square<KING>(strongSide));
Square strongPawn = normalize(pos, strongSide, pos.square<PAWN>(strongSide));
Square weakKing = normalize(pos, strongSide, pos.square<KING>(weakSide));
Color us = strongSide == pos.side_to_move() ? WHITE : BLACK;
if (!Bitbases::probe(strongKing, strongPawn, weakKing, us))
return VALUE_DRAW;
Value result = VALUE_KNOWN_WIN + PawnValueEg + Value(rank_of(strongPawn));
return strongSide == pos.side_to_move() ? result : -result;
}
/// KR vs KP. This is a somewhat tricky endgame to evaluate precisely without
/// a bitbase. The function below returns drawish scores when the pawn is
/// far advanced with support of the king, while the attacking king is far
/// away.
template<>
Value Endgame<KRKP>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, RookValueMg, 0));
assert(verify_material(pos, weakSide, VALUE_ZERO, 1));
Square strongKing = pos.square<KING>(strongSide);
Square weakKing = pos.square<KING>(weakSide);
Square strongRook = pos.square<ROOK>(strongSide);
Square weakPawn = pos.square<PAWN>(weakSide);
Square queeningSquare = make_square(file_of(weakPawn), relative_rank(weakSide, RANK_8));
Value result;
// If the stronger side's king is in front of the pawn, it's a win
if (forward_file_bb(strongSide, strongKing) & weakPawn)
result = RookValueEg - distance(strongKing, weakPawn);
// If the weaker side's king is too far from the pawn and the rook,
// it's a win.
else if ( distance(weakKing, weakPawn) >= 3 + (pos.side_to_move() == weakSide)
&& distance(weakKing, strongRook) >= 3)
result = RookValueEg - distance(strongKing, weakPawn);
// If the pawn is far advanced and supported by the defending king,
// the position is drawish
else if ( relative_rank(strongSide, weakKing) <= RANK_3
&& distance(weakKing, weakPawn) == 1
&& relative_rank(strongSide, strongKing) >= RANK_4
&& distance(strongKing, weakPawn) > 2 + (pos.side_to_move() == strongSide))
result = Value(80) - 8 * distance(strongKing, weakPawn);
else
result = Value(200) - 8 * ( distance(strongKing, weakPawn + pawn_push(weakSide))
- distance(weakKing, weakPawn + pawn_push(weakSide))
- distance(weakPawn, queeningSquare));
return strongSide == pos.side_to_move() ? result : -result;
}
/// KR vs KB. This is very simple, and always returns drawish scores. The
/// score is slightly bigger when the defending king is close to the edge.
template<>
Value Endgame<KRKB>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, RookValueMg, 0));
assert(verify_material(pos, weakSide, BishopValueMg, 0));
Value result = Value(push_to_edge(pos.square<KING>(weakSide)));
return strongSide == pos.side_to_move() ? result : -result;
}
/// KR vs KN. The attacking side has slightly better winning chances than
/// in KR vs KB, particularly if the king and the knight are far apart.
template<>
Value Endgame<KRKN>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, RookValueMg, 0));
assert(verify_material(pos, weakSide, KnightValueMg, 0));
Square weakKing = pos.square<KING>(weakSide);
Square weakKnight = pos.square<KNIGHT>(weakSide);
Value result = Value(push_to_edge(weakKing) + push_away(weakKing, weakKnight));
return strongSide == pos.side_to_move() ? result : -result;
}
/// KQ vs KP. In general, this is a win for the stronger side, but there are a
/// few important exceptions. A pawn on 7th rank and on the A,C,F or H files
/// with a king positioned next to it can be a draw, so in that case, we only
/// use the distance between the kings.
template<>
Value Endgame<KQKP>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, QueenValueMg, 0));
assert(verify_material(pos, weakSide, VALUE_ZERO, 1));
Square strongKing = pos.square<KING>(strongSide);
Square weakKing = pos.square<KING>(weakSide);
Square weakPawn = pos.square<PAWN>(weakSide);
Value result = Value(push_close(strongKing, weakKing));
if ( relative_rank(weakSide, weakPawn) != RANK_7
|| distance(weakKing, weakPawn) != 1
|| ((FileBBB | FileDBB | FileEBB | FileGBB) & weakPawn))
result += QueenValueEg - PawnValueEg;
return strongSide == pos.side_to_move() ? result : -result;
}
/// KQ vs KR. This is almost identical to KX vs K: we give the attacking
/// king a bonus for having the kings close together, and for forcing the
/// defending king towards the edge. If we also take care to avoid null move for
/// the defending side in the search, this is usually sufficient to win KQ vs KR.
template<>
Value Endgame<KQKR>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, QueenValueMg, 0));
assert(verify_material(pos, weakSide, RookValueMg, 0));
Square strongKing = pos.square<KING>(strongSide);
Square weakKing = pos.square<KING>(weakSide);
Value result = QueenValueEg
- RookValueEg
+ push_to_edge(weakKing)
+ push_close(strongKing, weakKing);
return strongSide == pos.side_to_move() ? result : -result;
}
/// KNN vs KP. Very drawish, but there are some mate opportunities if we can
/// press the weakSide King to a corner before the pawn advances too much.
template<>
Value Endgame<KNNKP>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, 2 * KnightValueMg, 0));
assert(verify_material(pos, weakSide, VALUE_ZERO, 1));
Square weakKing = pos.square<KING>(weakSide);
Square weakPawn = pos.square<PAWN>(weakSide);
Value result = PawnValueEg
+ 2 * push_to_edge(weakKing)
- 10 * relative_rank(weakSide, weakPawn);
return strongSide == pos.side_to_move() ? result : -result;
}
/// Some cases of trivial draws
template<> Value Endgame<KNNK>::operator()(const Position&) const { return VALUE_DRAW; }
/// KB and one or more pawns vs K. It checks for draws with rook pawns and
/// a bishop of the wrong color. If such a draw is detected, SCALE_FACTOR_DRAW
/// is returned. If not, the return value is SCALE_FACTOR_NONE, i.e. no scaling
/// will be used.
template<>
ScaleFactor Endgame<KBPsK>::operator()(const Position& pos) const {
assert(pos.non_pawn_material(strongSide) == BishopValueMg);
assert(pos.count<PAWN>(strongSide) >= 1);
// No assertions about the material of weakSide, because we want draws to
// be detected even when the weaker side has some pawns.
Bitboard strongPawns = pos.pieces(strongSide, PAWN);
Bitboard allPawns = pos.pieces(PAWN);
Square strongBishop = pos.square<BISHOP>(strongSide);
Square weakKing = pos.square<KING>(weakSide);
Square strongKing = pos.square<KING>(strongSide);
// All strongSide pawns are on a single rook file?
if (!(strongPawns & ~FileABB) || !(strongPawns & ~FileHBB))
{
Square queeningSquare = relative_square(strongSide, make_square(file_of(lsb(strongPawns)), RANK_8));
if ( opposite_colors(queeningSquare, strongBishop)
&& distance(queeningSquare, weakKing) <= 1)
return SCALE_FACTOR_DRAW;
}
// If all the pawns are on the same B or G file, then it's potentially a draw
if ((!(allPawns & ~FileBBB) || !(allPawns & ~FileGBB))
&& pos.non_pawn_material(weakSide) == 0
&& pos.count<PAWN>(weakSide) >= 1)
{
// Get the least advanced weakSide pawn
Square weakPawn = frontmost_sq(strongSide, pos.pieces(weakSide, PAWN));
// There's potential for a draw if our pawn is blocked on the 7th rank,
// the bishop cannot attack it or they only have one pawn left.
if ( relative_rank(strongSide, weakPawn) == RANK_7
&& (strongPawns & (weakPawn + pawn_push(weakSide)))
&& (opposite_colors(strongBishop, weakPawn) || !more_than_one(strongPawns)))
{
int strongKingDist = distance(weakPawn, strongKing);
int weakKingDist = distance(weakPawn, weakKing);
// It's a draw if the weak king is on its back two ranks, within 2
// squares of the blocking pawn and the strong king is not
// closer. (I think this rule only fails in practically
// unreachable positions such as 5k1K/6p1/6P1/8/8/3B4/8/8 w
// and positions where qsearch will immediately correct the
// problem such as 8/4k1p1/6P1/1K6/3B4/8/8/8 w).
if ( relative_rank(strongSide, weakKing) >= RANK_7
&& weakKingDist <= 2
&& weakKingDist <= strongKingDist)
return SCALE_FACTOR_DRAW;
}
}
return SCALE_FACTOR_NONE;
}
/// KQ vs KR and one or more pawns. It tests for fortress draws with a rook on
/// the third rank defended by a pawn.
template<>
ScaleFactor Endgame<KQKRPs>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, QueenValueMg, 0));
assert(pos.count<ROOK>(weakSide) == 1);
assert(pos.count<PAWN>(weakSide) >= 1);
Square strongKing = pos.square<KING>(strongSide);
Square weakKing = pos.square<KING>(weakSide);
Square weakRook = pos.square<ROOK>(weakSide);
if ( relative_rank(weakSide, weakKing) <= RANK_2
&& relative_rank(weakSide, strongKing) >= RANK_4
&& relative_rank(weakSide, weakRook) == RANK_3
&& ( pos.pieces(weakSide, PAWN)
& attacks_bb<KING>(weakKing)
& pawn_attacks_bb(strongSide, weakRook)))
return SCALE_FACTOR_DRAW;
return SCALE_FACTOR_NONE;
}
/// KRP vs KR. This function knows a handful of the most important classes of
/// drawn positions, but is far from perfect. It would probably be a good idea
/// to add more knowledge in the future.
///
/// It would also be nice to rewrite the actual code for this function,
/// which is mostly copied from Glaurung 1.x, and isn't very pretty.
template<>
ScaleFactor Endgame<KRPKR>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, RookValueMg, 1));
assert(verify_material(pos, weakSide, RookValueMg, 0));
// Assume strongSide is white and the pawn is on files A-D
Square strongKing = normalize(pos, strongSide, pos.square<KING>(strongSide));
Square strongRook = normalize(pos, strongSide, pos.square<ROOK>(strongSide));
Square strongPawn = normalize(pos, strongSide, pos.square<PAWN>(strongSide));
Square weakKing = normalize(pos, strongSide, pos.square<KING>(weakSide));
Square weakRook = normalize(pos, strongSide, pos.square<ROOK>(weakSide));
File pawnFile = file_of(strongPawn);
Rank pawnRank = rank_of(strongPawn);
Square queeningSquare = make_square(pawnFile, RANK_8);
int tempo = (pos.side_to_move() == strongSide);
// If the pawn is not too far advanced and the defending king defends the
// queening square, use the third-rank defence.
if ( pawnRank <= RANK_5
&& distance(weakKing, queeningSquare) <= 1
&& strongKing <= SQ_H5
&& (rank_of(weakRook) == RANK_6 || (pawnRank <= RANK_3 && rank_of(strongRook) != RANK_6)))
return SCALE_FACTOR_DRAW;
// The defending side saves a draw by checking from behind in case the pawn
// has advanced to the 6th rank with the king behind.
if ( pawnRank == RANK_6
&& distance(weakKing, queeningSquare) <= 1
&& rank_of(strongKing) + tempo <= RANK_6
&& (rank_of(weakRook) == RANK_1 || (!tempo && distance<File>(weakRook, strongPawn) >= 3)))
return SCALE_FACTOR_DRAW;
if ( pawnRank >= RANK_6
&& weakKing == queeningSquare
&& rank_of(weakRook) == RANK_1
&& (!tempo || distance(strongKing, strongPawn) >= 2))
return SCALE_FACTOR_DRAW;
// White pawn on a7 and rook on a8 is a draw if black's king is on g7 or h7
// and the black rook is behind the pawn.
if ( strongPawn == SQ_A7
&& strongRook == SQ_A8
&& (weakKing == SQ_H7 || weakKing == SQ_G7)
&& file_of(weakRook) == FILE_A
&& (rank_of(weakRook) <= RANK_3 || file_of(strongKing) >= FILE_D || rank_of(strongKing) <= RANK_5))
return SCALE_FACTOR_DRAW;
// If the defending king blocks the pawn and the attacking king is too far
// away, it's a draw.
if ( pawnRank <= RANK_5
&& weakKing == strongPawn + NORTH
&& distance(strongKing, strongPawn) - tempo >= 2
&& distance(strongKing, weakRook) - tempo >= 2)
return SCALE_FACTOR_DRAW;
// Pawn on the 7th rank supported by the rook from behind usually wins if the
// attacking king is closer to the queening square than the defending king,
// and the defending king cannot gain tempi by threatening the attacking rook.
if ( pawnRank == RANK_7
&& pawnFile != FILE_A
&& file_of(strongRook) == pawnFile
&& strongRook != queeningSquare
&& (distance(strongKing, queeningSquare) < distance(weakKing, queeningSquare) - 2 + tempo)
&& (distance(strongKing, queeningSquare) < distance(weakKing, strongRook) + tempo))
return ScaleFactor(SCALE_FACTOR_MAX - 2 * distance(strongKing, queeningSquare));
// Similar to the above, but with the pawn further back
if ( pawnFile != FILE_A
&& file_of(strongRook) == pawnFile
&& strongRook < strongPawn
&& (distance(strongKing, queeningSquare) < distance(weakKing, queeningSquare) - 2 + tempo)
&& (distance(strongKing, strongPawn + NORTH) < distance(weakKing, strongPawn + NORTH) - 2 + tempo)
&& ( distance(weakKing, strongRook) + tempo >= 3
|| ( distance(strongKing, queeningSquare) < distance(weakKing, strongRook) + tempo
&& (distance(strongKing, strongPawn + NORTH) < distance(weakKing, strongPawn) + tempo))))
return ScaleFactor( SCALE_FACTOR_MAX
- 8 * distance(strongPawn, queeningSquare)
- 2 * distance(strongKing, queeningSquare));
// If the pawn is not far advanced and the defending king is somewhere in
// the pawn's path, it's probably a draw.
if (pawnRank <= RANK_4 && weakKing > strongPawn)
{
if (file_of(weakKing) == file_of(strongPawn))
return ScaleFactor(10);
if ( distance<File>(weakKing, strongPawn) == 1
&& distance(strongKing, weakKing) > 2)
return ScaleFactor(24 - 2 * distance(strongKing, weakKing));
}
return SCALE_FACTOR_NONE;
}
template<>
ScaleFactor Endgame<KRPKB>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, RookValueMg, 1));
assert(verify_material(pos, weakSide, BishopValueMg, 0));
// Test for a rook pawn
if (pos.pieces(PAWN) & (FileABB | FileHBB))
{
Square weakKing = pos.square<KING>(weakSide);
Square weakBishop = pos.square<BISHOP>(weakSide);
Square strongKing = pos.square<KING>(strongSide);
Square strongPawn = pos.square<PAWN>(strongSide);
Rank pawnRank = relative_rank(strongSide, strongPawn);
Direction push = pawn_push(strongSide);
// If the pawn is on the 5th rank and the pawn (currently) is on
// the same color square as the bishop then there is a chance of
// a fortress. Depending on the king position give a moderate
// reduction or a stronger one if the defending king is near the
// corner but not trapped there.
if (pawnRank == RANK_5 && !opposite_colors(weakBishop, strongPawn))
{
int d = distance(strongPawn + 3 * push, weakKing);
if (d <= 2 && !(d == 0 && weakKing == strongKing + 2 * push))
return ScaleFactor(24);
else
return ScaleFactor(48);
}
// When the pawn has moved to the 6th rank we can be fairly sure
// it's drawn if the bishop attacks the square in front of the
// pawn from a reasonable distance and the defending king is near
// the corner
if ( pawnRank == RANK_6
&& distance(strongPawn + 2 * push, weakKing) <= 1
&& (attacks_bb<BISHOP>(weakBishop) & (strongPawn + push))
&& distance<File>(weakBishop, strongPawn) >= 2)
return ScaleFactor(8);
}
return SCALE_FACTOR_NONE;
}
/// KRPP vs KRP. There is just a single rule: if the stronger side has no passed
/// pawns and the defending king is actively placed, the position is drawish.
template<>
ScaleFactor Endgame<KRPPKRP>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, RookValueMg, 2));
assert(verify_material(pos, weakSide, RookValueMg, 1));
Square strongPawn1 = lsb(pos.pieces(strongSide, PAWN));
Square strongPawn2 = msb(pos.pieces(strongSide, PAWN));
Square weakKing = pos.square<KING>(weakSide);
// Does the stronger side have a passed pawn?
if (pos.pawn_passed(strongSide, strongPawn1) || pos.pawn_passed(strongSide, strongPawn2))
return SCALE_FACTOR_NONE;
Rank pawnRank = std::max(relative_rank(strongSide, strongPawn1), relative_rank(strongSide, strongPawn2));
if ( distance<File>(weakKing, strongPawn1) <= 1
&& distance<File>(weakKing, strongPawn2) <= 1
&& relative_rank(strongSide, weakKing) > pawnRank)
{
assert(pawnRank > RANK_1 && pawnRank < RANK_7);
return ScaleFactor(7 * pawnRank);
}
return SCALE_FACTOR_NONE;
}
/// K and two or more pawns vs K. There is just a single rule here: if all pawns
/// are on the same rook file and are blocked by the defending king, it's a draw.
template<>
ScaleFactor Endgame<KPsK>::operator()(const Position& pos) const {
assert(pos.non_pawn_material(strongSide) == VALUE_ZERO);
assert(pos.count<PAWN>(strongSide) >= 2);
assert(verify_material(pos, weakSide, VALUE_ZERO, 0));
Square weakKing = pos.square<KING>(weakSide);
Bitboard strongPawns = pos.pieces(strongSide, PAWN);
// If all pawns are ahead of the king on a single rook file, it's a draw.
if ( !(strongPawns & ~(FileABB | FileHBB))
&& !(strongPawns & ~passed_pawn_span(weakSide, weakKing)))
return SCALE_FACTOR_DRAW;
return SCALE_FACTOR_NONE;
}
/// KBP vs KB. There are two rules: if the defending king is somewhere along the
/// path of the pawn, and the square of the king is not of the same color as the
/// stronger side's bishop, it's a draw. If the two bishops have opposite color,
/// it's almost always a draw.
template<>
ScaleFactor Endgame<KBPKB>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, BishopValueMg, 1));
assert(verify_material(pos, weakSide, BishopValueMg, 0));
Square strongPawn = pos.square<PAWN>(strongSide);
Square strongBishop = pos.square<BISHOP>(strongSide);
Square weakBishop = pos.square<BISHOP>(weakSide);
Square weakKing = pos.square<KING>(weakSide);
// Case 1: Defending king blocks the pawn, and cannot be driven away
if ( (forward_file_bb(strongSide, strongPawn) & weakKing)
&& ( opposite_colors(weakKing, strongBishop)
|| relative_rank(strongSide, weakKing) <= RANK_6))
return SCALE_FACTOR_DRAW;
// Case 2: Opposite colored bishops
if (opposite_colors(strongBishop, weakBishop))
return SCALE_FACTOR_DRAW;
return SCALE_FACTOR_NONE;
}
/// KBPP vs KB. It detects a few basic draws with opposite-colored bishops
template<>
ScaleFactor Endgame<KBPPKB>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, BishopValueMg, 2));
assert(verify_material(pos, weakSide, BishopValueMg, 0));
Square strongBishop = pos.square<BISHOP>(strongSide);
Square weakBishop = pos.square<BISHOP>(weakSide);
if (!opposite_colors(strongBishop, weakBishop))
return SCALE_FACTOR_NONE;
Square weakKing = pos.square<KING>(weakSide);
Square strongPawn1 = lsb(pos.pieces(strongSide, PAWN));
Square strongPawn2 = msb(pos.pieces(strongSide, PAWN));
Square blockSq1, blockSq2;
if (relative_rank(strongSide, strongPawn1) > relative_rank(strongSide, strongPawn2))
{
blockSq1 = strongPawn1 + pawn_push(strongSide);
blockSq2 = make_square(file_of(strongPawn2), rank_of(strongPawn1));
}
else
{
blockSq1 = strongPawn2 + pawn_push(strongSide);
blockSq2 = make_square(file_of(strongPawn1), rank_of(strongPawn2));
}
switch (distance<File>(strongPawn1, strongPawn2))
{
case 0:
// Both pawns are on the same file. It's an easy draw if the defender firmly
// controls some square in the frontmost pawn's path.
if ( file_of(weakKing) == file_of(blockSq1)
&& relative_rank(strongSide, weakKing) >= relative_rank(strongSide, blockSq1)
&& opposite_colors(weakKing, strongBishop))
return SCALE_FACTOR_DRAW;
else
return SCALE_FACTOR_NONE;
case 1:
// Pawns on adjacent files. It's a draw if the defender firmly controls the
// square in front of the frontmost pawn's path, and the square diagonally
// behind this square on the file of the other pawn.
if ( weakKing == blockSq1
&& opposite_colors(weakKing, strongBishop)
&& ( weakBishop == blockSq2
|| (attacks_bb<BISHOP>(blockSq2, pos.pieces()) & pos.pieces(weakSide, BISHOP))
|| distance<Rank>(strongPawn1, strongPawn2) >= 2))
return SCALE_FACTOR_DRAW;
else if ( weakKing == blockSq2
&& opposite_colors(weakKing, strongBishop)
&& ( weakBishop == blockSq1
|| (attacks_bb<BISHOP>(blockSq1, pos.pieces()) & pos.pieces(weakSide, BISHOP))))
return SCALE_FACTOR_DRAW;
else
return SCALE_FACTOR_NONE;
default:
// The pawns are not on the same file or adjacent files. No scaling.
return SCALE_FACTOR_NONE;
}
}
/// KBP vs KN. There is a single rule: if the defending king is somewhere along
/// the path of the pawn, and the square of the king is not of the same color as
/// the stronger side's bishop, it's a draw.
template<>
ScaleFactor Endgame<KBPKN>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, BishopValueMg, 1));
assert(verify_material(pos, weakSide, KnightValueMg, 0));
Square strongPawn = pos.square<PAWN>(strongSide);
Square strongBishop = pos.square<BISHOP>(strongSide);
Square weakKing = pos.square<KING>(weakSide);
if ( file_of(weakKing) == file_of(strongPawn)
&& relative_rank(strongSide, strongPawn) < relative_rank(strongSide, weakKing)
&& ( opposite_colors(weakKing, strongBishop)
|| relative_rank(strongSide, weakKing) <= RANK_6))
return SCALE_FACTOR_DRAW;
return SCALE_FACTOR_NONE;
}
/// KP vs KP. This is done by removing the weakest side's pawn and probing the
/// KP vs K bitbase: if the weakest side has a draw without the pawn, it probably
/// has at least a draw with the pawn as well. The exception is when the stronger
/// side's pawn is far advanced and not on a rook file; in this case it is often
/// possible to win (e.g. 8/4k3/3p4/3P4/6K1/8/8/8 w - - 0 1).
template<>
ScaleFactor Endgame<KPKP>::operator()(const Position& pos) const {
assert(verify_material(pos, strongSide, VALUE_ZERO, 1));
assert(verify_material(pos, weakSide, VALUE_ZERO, 1));
// Assume strongSide is white and the pawn is on files A-D
Square strongKing = normalize(pos, strongSide, pos.square<KING>(strongSide));
Square weakKing = normalize(pos, strongSide, pos.square<KING>(weakSide));
Square strongPawn = normalize(pos, strongSide, pos.square<PAWN>(strongSide));
Color us = strongSide == pos.side_to_move() ? WHITE : BLACK;
// If the pawn has advanced to the fifth rank or further, and is not a
// rook pawn, it's too dangerous to assume that it's at least a draw.
if (rank_of(strongPawn) >= RANK_5 && file_of(strongPawn) != FILE_A)
return SCALE_FACTOR_NONE;
// Probe the KPK bitbase with the weakest side's pawn removed. If it's a draw,
// it's probably at least a draw even with the pawn.
return Bitbases::probe(strongKing, strongPawn, weakKing, us) ? SCALE_FACTOR_NONE : SCALE_FACTOR_DRAW;
}
} // namespace Stockfish

126
src/Stockfish/src/endgame.h Normal file
View file

@ -0,0 +1,126 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef ENDGAME_H_INCLUDED
#define ENDGAME_H_INCLUDED
#include <memory>
#include <string>
#include <type_traits>
#include <unordered_map>
#include <utility>
#include "position.h"
#include "types.h"
namespace Stockfish {
/// EndgameCode lists all supported endgame functions by corresponding codes
enum EndgameCode {
EVALUATION_FUNCTIONS,
KNNK, // KNN vs K
KNNKP, // KNN vs KP
KXK, // Generic "mate lone king" eval
KBNK, // KBN vs K
KPK, // KP vs K
KRKP, // KR vs KP
KRKB, // KR vs KB
KRKN, // KR vs KN
KQKP, // KQ vs KP
KQKR, // KQ vs KR
SCALING_FUNCTIONS,
KBPsK, // KB and pawns vs K
KQKRPs, // KQ vs KR and pawns
KRPKR, // KRP vs KR
KRPKB, // KRP vs KB
KRPPKRP, // KRPP vs KRP
KPsK, // K and pawns vs K
KBPKB, // KBP vs KB
KBPPKB, // KBPP vs KB
KBPKN, // KBP vs KN
KPKP // KP vs KP
};
/// Endgame functions can be of two types depending on whether they return a
/// Value or a ScaleFactor.
template<EndgameCode E> using
eg_type = typename std::conditional<(E < SCALING_FUNCTIONS), Value, ScaleFactor>::type;
/// Base and derived functors for endgame evaluation and scaling functions
template<typename T>
struct EndgameBase {
explicit EndgameBase(Color c) : strongSide(c), weakSide(~c) {}
virtual ~EndgameBase() = default;
virtual T operator()(const Position&) const = 0;
const Color strongSide, weakSide;
};
template<EndgameCode E, typename T = eg_type<E>>
struct Endgame : public EndgameBase<T> {
explicit Endgame(Color c) : EndgameBase<T>(c) {}
T operator()(const Position&) const override;
};
/// The Endgames namespace handles the pointers to endgame evaluation and scaling
/// base objects in two std::map. We use polymorphism to invoke the actual
/// endgame function by calling its virtual operator().
namespace Endgames {
template<typename T> using Ptr = std::unique_ptr<EndgameBase<T>>;
template<typename T> using Map = std::unordered_map<Key, Ptr<T>>;
extern std::pair<Map<Value>, Map<ScaleFactor>> maps;
void init();
template<typename T>
Map<T>& map() {
return std::get<std::is_same<T, ScaleFactor>::value>(maps);
}
template<EndgameCode E, typename T = eg_type<E>>
void add(const std::string& code) {
StateInfo st;
map<T>()[Position().set(code, WHITE, &st).material_key()] = Ptr<T>(new Endgame<E>(WHITE));
map<T>()[Position().set(code, BLACK, &st).material_key()] = Ptr<T>(new Endgame<E>(BLACK));
}
template<typename T>
const EndgameBase<T>* probe(Key key) {
auto it = map<T>().find(key);
return it != map<T>().end() ? it->second.get() : nullptr;
}
}
} // namespace Stockfish
#endif // #ifndef ENDGAME_H_INCLUDED

1158
src/Stockfish/src/evaluate.cpp Normal file
View file

File diff suppressed because it is too large Load diff

55
src/Stockfish/src/evaluate.h Normal file
View file

@ -0,0 +1,55 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef EVALUATE_H_INCLUDED
#define EVALUATE_H_INCLUDED
#include <string>
#include <optional>
#include "types.h"
namespace Stockfish {
class Position;
namespace Eval {
std::string trace(Position& pos);
Value evaluate(const Position& pos);
extern bool useNNUE;
extern std::string currentEvalFileName;
// The default net name MUST follow the format nn-[SHA256 first 12 digits].nnue
// for the build process (profile-build and fishtest) to work. Do not change the
// name of the macro, as it is used in the Makefile.
#define EvalFileDefaultName "nn-5af11540bbfe.nnue"
namespace NNUE {
void init();
void verify();
} // namespace NNUE
} // namespace Eval
} // namespace Stockfish
#endif // #ifndef EVALUATE_H_INCLUDED

26
src/Stockfish/src/incbin/UNLICENCE Normal file
View file

@ -0,0 +1,26 @@
The file "incbin.h" is free and unencumbered software released into
the public domain by Dale Weiler, see:
<https://github.com/graphitemaster/incbin>
Anyone is free to copy, modify, publish, use, compile, sell, or
distribute this software, either in source code form or as a compiled
binary, for any purpose, commercial or non-commercial, and by any
means.
In jurisdictions that recognize copyright laws, the author or authors
of this software dedicate any and all copyright interest in the
software to the public domain. We make this dedication for the benefit
of the public at large and to the detriment of our heirs and
successors. We intend this dedication to be an overt act of
relinquishment in perpetuity of all present and future rights to this
software under copyright law.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR
OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
OTHER DEALINGS IN THE SOFTWARE.
For more information, please refer to <http://unlicense.org/>

368
src/Stockfish/src/incbin/incbin.h Normal file
View file

@ -0,0 +1,368 @@
/**
* @file incbin.h
* @author Dale Weiler
* @brief Utility for including binary files
*
* Facilities for including binary files into the current translation unit and
* making use from them externally in other translation units.
*/
#ifndef INCBIN_HDR
#define INCBIN_HDR
#include <limits.h>
#if defined(__AVX512BW__) || \
defined(__AVX512CD__) || \
defined(__AVX512DQ__) || \
defined(__AVX512ER__) || \
defined(__AVX512PF__) || \
defined(__AVX512VL__) || \
defined(__AVX512F__)
# define INCBIN_ALIGNMENT_INDEX 6
#elif defined(__AVX__) || \
defined(__AVX2__)
# define INCBIN_ALIGNMENT_INDEX 5
#elif defined(__SSE__) || \
defined(__SSE2__) || \
defined(__SSE3__) || \
defined(__SSSE3__) || \
defined(__SSE4_1__) || \
defined(__SSE4_2__) || \
defined(__neon__)
# define INCBIN_ALIGNMENT_INDEX 4
#elif ULONG_MAX != 0xffffffffu
# define INCBIN_ALIGNMENT_INDEX 3
# else
# define INCBIN_ALIGNMENT_INDEX 2
#endif
/* Lookup table of (1 << n) where `n' is `INCBIN_ALIGNMENT_INDEX' */
#define INCBIN_ALIGN_SHIFT_0 1
#define INCBIN_ALIGN_SHIFT_1 2
#define INCBIN_ALIGN_SHIFT_2 4
#define INCBIN_ALIGN_SHIFT_3 8
#define INCBIN_ALIGN_SHIFT_4 16
#define INCBIN_ALIGN_SHIFT_5 32
#define INCBIN_ALIGN_SHIFT_6 64
/* Actual alignment value */
#define INCBIN_ALIGNMENT \
INCBIN_CONCATENATE( \
INCBIN_CONCATENATE(INCBIN_ALIGN_SHIFT, _), \
INCBIN_ALIGNMENT_INDEX)
/* Stringize */
#define INCBIN_STR(X) \
#X
#define INCBIN_STRINGIZE(X) \
INCBIN_STR(X)
/* Concatenate */
#define INCBIN_CAT(X, Y) \
X ## Y
#define INCBIN_CONCATENATE(X, Y) \
INCBIN_CAT(X, Y)
/* Deferred macro expansion */
#define INCBIN_EVAL(X) \
X
#define INCBIN_INVOKE(N, ...) \
INCBIN_EVAL(N(__VA_ARGS__))
/* Green Hills uses a different directive for including binary data */
#if defined(__ghs__)
# if (__ghs_asm == 2)
# define INCBIN_MACRO ".file"
/* Or consider the ".myrawdata" entry in the ld file */
# else
# define INCBIN_MACRO "\tINCBIN"
# endif
#else
# define INCBIN_MACRO ".incbin"
#endif
#ifndef _MSC_VER
# define INCBIN_ALIGN \
__attribute__((aligned(INCBIN_ALIGNMENT)))
#else
# define INCBIN_ALIGN __declspec(align(INCBIN_ALIGNMENT))
#endif
#if defined(__arm__) || /* GNU C and RealView */ \
defined(__arm) || /* Diab */ \
defined(_ARM) /* ImageCraft */
# define INCBIN_ARM
#endif
#ifdef __GNUC__
/* Utilize .balign where supported */
# define INCBIN_ALIGN_HOST ".balign " INCBIN_STRINGIZE(INCBIN_ALIGNMENT) "\n"
# define INCBIN_ALIGN_BYTE ".balign 1\n"
#elif defined(INCBIN_ARM)
/*
* On arm assemblers, the alignment value is calculated as (1 << n) where `n' is
* the shift count. This is the value passed to `.align'
*/
# define INCBIN_ALIGN_HOST ".align " INCBIN_STRINGIZE(INCBIN_ALIGNMENT_INDEX) "\n"
# define INCBIN_ALIGN_BYTE ".align 0\n"
#else
/* We assume other inline assembler's treat `.align' as `.balign' */
# define INCBIN_ALIGN_HOST ".align " INCBIN_STRINGIZE(INCBIN_ALIGNMENT) "\n"
# define INCBIN_ALIGN_BYTE ".align 1\n"
#endif
/* INCBIN_CONST is used by incbin.c generated files */
#if defined(__cplusplus)
# define INCBIN_EXTERNAL extern "C"
# define INCBIN_CONST extern const
#else
# define INCBIN_EXTERNAL extern
# define INCBIN_CONST const
#endif
/**
* @brief Optionally override the linker section into which data is emitted.
*
* @warning If you use this facility, you'll have to deal with platform-specific linker output
* section naming on your own
*
* Overriding the default linker output section, e.g for esp8266/Arduino:
* @code
* #define INCBIN_OUTPUT_SECTION ".irom.text"
* #include "incbin.h"
* INCBIN(Foo, "foo.txt");
* // Data is emitted into program memory that never gets copied to RAM
* @endcode
*/
#if !defined(INCBIN_OUTPUT_SECTION)
# if defined(__APPLE__)
# define INCBIN_OUTPUT_SECTION ".const_data"
# else
# define INCBIN_OUTPUT_SECTION ".rodata"
# endif
#endif
#if defined(__APPLE__)
/* The directives are different for Apple branded compilers */
# define INCBIN_SECTION INCBIN_OUTPUT_SECTION "\n"
# define INCBIN_GLOBAL(NAME) ".globl " INCBIN_MANGLE INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME "\n"
# define INCBIN_INT ".long "
# define INCBIN_MANGLE "_"
# define INCBIN_BYTE ".byte "
# define INCBIN_TYPE(...)
#else
# define INCBIN_SECTION ".section " INCBIN_OUTPUT_SECTION "\n"
# define INCBIN_GLOBAL(NAME) ".global " INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME "\n"
# if defined(__ghs__)
# define INCBIN_INT ".word "
# else
# define INCBIN_INT ".int "
# endif
# if defined(__USER_LABEL_PREFIX__)
# define INCBIN_MANGLE INCBIN_STRINGIZE(__USER_LABEL_PREFIX__)
# else
# define INCBIN_MANGLE ""
# endif
# if defined(INCBIN_ARM)
/* On arm assemblers, `@' is used as a line comment token */
# define INCBIN_TYPE(NAME) ".type " INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME ", %object\n"
# elif defined(__MINGW32__) || defined(__MINGW64__)
/* Mingw doesn't support this directive either */
# define INCBIN_TYPE(NAME)
# else
/* It's safe to use `@' on other architectures */
# define INCBIN_TYPE(NAME) ".type " INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME ", @object\n"
# endif
# define INCBIN_BYTE ".byte "
#endif
/* List of style types used for symbol names */
#define INCBIN_STYLE_CAMEL 0
#define INCBIN_STYLE_SNAKE 1
/**
* @brief Specify the prefix to use for symbol names.
*
* By default this is `g', producing symbols of the form:
* @code
* #include "incbin.h"
* INCBIN(Foo, "foo.txt");
*
* // Now you have the following symbols:
* // const unsigned char gFooData[];
* // const unsigned char *const gFooEnd;
* // const unsigned int gFooSize;
* @endcode
*
* If however you specify a prefix before including: e.g:
* @code
* #define INCBIN_PREFIX incbin
* #include "incbin.h"
* INCBIN(Foo, "foo.txt");
*
* // Now you have the following symbols instead:
* // const unsigned char incbinFooData[];
* // const unsigned char *const incbinFooEnd;
* // const unsigned int incbinFooSize;
* @endcode
*/
#if !defined(INCBIN_PREFIX)
# define INCBIN_PREFIX g
#endif
/**
* @brief Specify the style used for symbol names.
*
* Possible options are
* - INCBIN_STYLE_CAMEL "CamelCase"
* - INCBIN_STYLE_SNAKE "snake_case"
*
* Default option is *INCBIN_STYLE_CAMEL* producing symbols of the form:
* @code
* #include "incbin.h"
* INCBIN(Foo, "foo.txt");
*
* // Now you have the following symbols:
* // const unsigned char <prefix>FooData[];
* // const unsigned char *const <prefix>FooEnd;
* // const unsigned int <prefix>FooSize;
* @endcode
*
* If however you specify a style before including: e.g:
* @code
* #define INCBIN_STYLE INCBIN_STYLE_SNAKE
* #include "incbin.h"
* INCBIN(foo, "foo.txt");
*
* // Now you have the following symbols:
* // const unsigned char <prefix>foo_data[];
* // const unsigned char *const <prefix>foo_end;
* // const unsigned int <prefix>foo_size;
* @endcode
*/
#if !defined(INCBIN_STYLE)
# define INCBIN_STYLE INCBIN_STYLE_CAMEL
#endif
/* Style lookup tables */
#define INCBIN_STYLE_0_DATA Data
#define INCBIN_STYLE_0_END End
#define INCBIN_STYLE_0_SIZE Size
#define INCBIN_STYLE_1_DATA _data
#define INCBIN_STYLE_1_END _end
#define INCBIN_STYLE_1_SIZE _size
/* Style lookup: returning identifier */
#define INCBIN_STYLE_IDENT(TYPE) \
INCBIN_CONCATENATE( \
INCBIN_STYLE_, \
INCBIN_CONCATENATE( \
INCBIN_EVAL(INCBIN_STYLE), \
INCBIN_CONCATENATE(_, TYPE)))
/* Style lookup: returning string literal */
#define INCBIN_STYLE_STRING(TYPE) \
INCBIN_STRINGIZE( \
INCBIN_STYLE_IDENT(TYPE)) \
/* Generate the global labels by indirectly invoking the macro with our style
* type and concatenating the name against them. */
#define INCBIN_GLOBAL_LABELS(NAME, TYPE) \
INCBIN_INVOKE( \
INCBIN_GLOBAL, \
INCBIN_CONCATENATE( \
NAME, \
INCBIN_INVOKE( \
INCBIN_STYLE_IDENT, \
TYPE))) \
INCBIN_INVOKE( \
INCBIN_TYPE, \
INCBIN_CONCATENATE( \
NAME, \
INCBIN_INVOKE( \
INCBIN_STYLE_IDENT, \
TYPE)))
/**
* @brief Externally reference binary data included in another translation unit.
*
* Produces three external symbols that reference the binary data included in
* another translation unit.
*
* The symbol names are a concatenation of `INCBIN_PREFIX' before *NAME*; with
* "Data", as well as "End" and "Size" after. An example is provided below.
*
* @param NAME The name given for the binary data
*
* @code
* INCBIN_EXTERN(Foo);
*
* // Now you have the following symbols:
* // extern const unsigned char <prefix>FooData[];
* // extern const unsigned char *const <prefix>FooEnd;
* // extern const unsigned int <prefix>FooSize;
* @endcode
*/
#define INCBIN_EXTERN(NAME) \
INCBIN_EXTERNAL const INCBIN_ALIGN unsigned char \
INCBIN_CONCATENATE( \
INCBIN_CONCATENATE(INCBIN_PREFIX, NAME), \
INCBIN_STYLE_IDENT(DATA))[]; \
INCBIN_EXTERNAL const INCBIN_ALIGN unsigned char *const \
INCBIN_CONCATENATE( \
INCBIN_CONCATENATE(INCBIN_PREFIX, NAME), \
INCBIN_STYLE_IDENT(END)); \
INCBIN_EXTERNAL const unsigned int \
INCBIN_CONCATENATE( \
INCBIN_CONCATENATE(INCBIN_PREFIX, NAME), \
INCBIN_STYLE_IDENT(SIZE))
/**
* @brief Include a binary file into the current translation unit.
*
* Includes a binary file into the current translation unit, producing three symbols
* for objects that encode the data and size respectively.
*
* The symbol names are a concatenation of `INCBIN_PREFIX' before *NAME*; with
* "Data", as well as "End" and "Size" after. An example is provided below.
*
* @param NAME The name to associate with this binary data (as an identifier.)
* @param FILENAME The file to include (as a string literal.)
*
* @code
* INCBIN(Icon, "icon.png");
*
* // Now you have the following symbols:
* // const unsigned char <prefix>IconData[];
* // const unsigned char *const <prefix>IconEnd;
* // const unsigned int <prefix>IconSize;
* @endcode
*
* @warning This must be used in global scope
* @warning The identifiers may be different if INCBIN_STYLE is not default
*
* To externally reference the data included by this in another translation unit
* please @see INCBIN_EXTERN.
*/
#ifdef _MSC_VER
#define INCBIN(NAME, FILENAME) \
INCBIN_EXTERN(NAME)
#else
#define INCBIN(NAME, FILENAME) \
__asm__(INCBIN_SECTION \
INCBIN_GLOBAL_LABELS(NAME, DATA) \
INCBIN_ALIGN_HOST \
INCBIN_MANGLE INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME INCBIN_STYLE_STRING(DATA) ":\n" \
INCBIN_MACRO " \"" FILENAME "\"\n" \
INCBIN_GLOBAL_LABELS(NAME, END) \
INCBIN_ALIGN_BYTE \
INCBIN_MANGLE INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME INCBIN_STYLE_STRING(END) ":\n" \
INCBIN_BYTE "1\n" \
INCBIN_GLOBAL_LABELS(NAME, SIZE) \
INCBIN_ALIGN_HOST \
INCBIN_MANGLE INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME INCBIN_STYLE_STRING(SIZE) ":\n" \
INCBIN_INT INCBIN_MANGLE INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME INCBIN_STYLE_STRING(END) " - " \
INCBIN_MANGLE INCBIN_STRINGIZE(INCBIN_PREFIX) #NAME INCBIN_STYLE_STRING(DATA) "\n" \
INCBIN_ALIGN_HOST \
".text\n" \
); \
INCBIN_EXTERN(NAME)
#endif
#endif

53
src/Stockfish/src/main.cpp Normal file
View file

@ -0,0 +1,53 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <iostream>
#include "bitboard.h"
#include "endgame.h"
#include "position.h"
#include "psqt.h"
#include "search.h"
#include "syzygy/tbprobe.h"
#include "thread.h"
#include "tt.h"
#include "uci.h"
using namespace Stockfish;
int main(int argc, char* argv[]) {
fakeout << engine_info() << fakeendl;
CommandLine::init(argc, argv);
UCI::init(Options);
Tune::init();
PSQT::init();
Bitboards::init();
Position::init();
Bitbases::init();
Endgames::init();
Threads.set(size_t(Options["Threads"]));
Search::clear(); // After threads are up
Eval::NNUE::init();
UCI::loop(argc, argv);
Threads.set(0);
return 0;
}

229
src/Stockfish/src/material.cpp Normal file
View file

@ -0,0 +1,229 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cassert>
#include <cstring> // For std::memset
#include "material.h"
#include "thread.h"
using namespace std;
namespace Stockfish {
namespace {
#define S(mg, eg) make_score(mg, eg)
// Polynomial material imbalance parameters
// One Score parameter for each pair (our piece, another of our pieces)
constexpr Score QuadraticOurs[][PIECE_TYPE_NB] = {
// OUR PIECE 2
// bishop pair pawn knight bishop rook queen
{S(1419, 1455) }, // Bishop pair
{S( 101, 28), S( 37, 39) }, // Pawn
{S( 57, 64), S(249, 187), S(-49, -62) }, // Knight OUR PIECE 1
{S( 0, 0), S(118, 137), S( 10, 27), S( 0, 0) }, // Bishop
{S( -63, -68), S( -5, 3), S(100, 81), S(132, 118), S(-246, -244) }, // Rook
{S(-210, -211), S( 37, 14), S(147, 141), S(161, 105), S(-158, -174), S(-9,-31) } // Queen
};
// One Score parameter for each pair (our piece, their piece)
constexpr Score QuadraticTheirs[][PIECE_TYPE_NB] = {
// THEIR PIECE
// bishop pair pawn knight bishop rook queen
{ }, // Bishop pair
{S( 33, 30) }, // Pawn
{S( 46, 18), S(106, 84) }, // Knight OUR PIECE
{S( 75, 35), S( 59, 44), S( 60, 15) }, // Bishop
{S( 26, 35), S( 6, 22), S( 38, 39), S(-12, -2) }, // Rook
{S( 97, 93), S(100, 163), S(-58, -91), S(112, 192), S(276, 225) } // Queen
};
#undef S
// Endgame evaluation and scaling functions are accessed directly and not through
// the function maps because they correspond to more than one material hash key.
Endgame<KXK> EvaluateKXK[] = { Endgame<KXK>(WHITE), Endgame<KXK>(BLACK) };
Endgame<KBPsK> ScaleKBPsK[] = { Endgame<KBPsK>(WHITE), Endgame<KBPsK>(BLACK) };
Endgame<KQKRPs> ScaleKQKRPs[] = { Endgame<KQKRPs>(WHITE), Endgame<KQKRPs>(BLACK) };
Endgame<KPsK> ScaleKPsK[] = { Endgame<KPsK>(WHITE), Endgame<KPsK>(BLACK) };
Endgame<KPKP> ScaleKPKP[] = { Endgame<KPKP>(WHITE), Endgame<KPKP>(BLACK) };
// Helper used to detect a given material distribution
bool is_KXK(const Position& pos, Color us) {
return !more_than_one(pos.pieces(~us))
&& pos.non_pawn_material(us) >= RookValueMg;
}
bool is_KBPsK(const Position& pos, Color us) {
return pos.non_pawn_material(us) == BishopValueMg
&& pos.count<PAWN>(us) >= 1;
}
bool is_KQKRPs(const Position& pos, Color us) {
return !pos.count<PAWN>(us)
&& pos.non_pawn_material(us) == QueenValueMg
&& pos.count<ROOK>(~us) == 1
&& pos.count<PAWN>(~us) >= 1;
}
/// imbalance() calculates the imbalance by comparing the piece count of each
/// piece type for both colors.
template<Color Us>
Score imbalance(const int pieceCount[][PIECE_TYPE_NB]) {
constexpr Color Them = ~Us;
Score bonus = SCORE_ZERO;
// Second-degree polynomial material imbalance, by Tord Romstad
for (int pt1 = NO_PIECE_TYPE; pt1 <= QUEEN; ++pt1)
{
if (!pieceCount[Us][pt1])
continue;
int v = QuadraticOurs[pt1][pt1] * pieceCount[Us][pt1];
for (int pt2 = NO_PIECE_TYPE; pt2 < pt1; ++pt2)
v += QuadraticOurs[pt1][pt2] * pieceCount[Us][pt2]
+ QuadraticTheirs[pt1][pt2] * pieceCount[Them][pt2];
bonus += pieceCount[Us][pt1] * v;
}
return bonus;
}
} // namespace
namespace Material {
/// Material::probe() looks up the current position's material configuration in
/// the material hash table. It returns a pointer to the Entry if the position
/// is found. Otherwise a new Entry is computed and stored there, so we don't
/// have to recompute all when the same material configuration occurs again.
Entry* probe(const Position& pos) {
Key key = pos.material_key();
Entry* e = pos.this_thread()->materialTable[key];
if (e->key == key)
return e;
std::memset(e, 0, sizeof(Entry));
e->key = key;
e->factor[WHITE] = e->factor[BLACK] = (uint8_t)SCALE_FACTOR_NORMAL;
Value npm_w = pos.non_pawn_material(WHITE);
Value npm_b = pos.non_pawn_material(BLACK);
Value npm = std::clamp(npm_w + npm_b, EndgameLimit, MidgameLimit);
// Map total non-pawn material into [PHASE_ENDGAME, PHASE_MIDGAME]
e->gamePhase = Phase(((npm - EndgameLimit) * PHASE_MIDGAME) / (MidgameLimit - EndgameLimit));
// Let's look if we have a specialized evaluation function for this particular
// material configuration. Firstly we look for a fixed configuration one, then
// for a generic one if the previous search failed.
if ((e->evaluationFunction = Endgames::probe<Value>(key)) != nullptr)
return e;
for (Color c : { WHITE, BLACK })
if (is_KXK(pos, c))
{
e->evaluationFunction = &EvaluateKXK[c];
return e;
}
// OK, we didn't find any special evaluation function for the current material
// configuration. Is there a suitable specialized scaling function?
const auto* sf = Endgames::probe<ScaleFactor>(key);
if (sf)
{
e->scalingFunction[sf->strongSide] = sf; // Only strong color assigned
return e;
}
// We didn't find any specialized scaling function, so fall back on generic
// ones that refer to more than one material distribution. Note that in this
// case we don't return after setting the function.
for (Color c : { WHITE, BLACK })
{
if (is_KBPsK(pos, c))
e->scalingFunction[c] = &ScaleKBPsK[c];
else if (is_KQKRPs(pos, c))
e->scalingFunction[c] = &ScaleKQKRPs[c];
}
if (npm_w + npm_b == VALUE_ZERO && pos.pieces(PAWN)) // Only pawns on the board
{
if (!pos.count<PAWN>(BLACK))
{
assert(pos.count<PAWN>(WHITE) >= 2);
e->scalingFunction[WHITE] = &ScaleKPsK[WHITE];
}
else if (!pos.count<PAWN>(WHITE))
{
assert(pos.count<PAWN>(BLACK) >= 2);
e->scalingFunction[BLACK] = &ScaleKPsK[BLACK];
}
else if (pos.count<PAWN>(WHITE) == 1 && pos.count<PAWN>(BLACK) == 1)
{
// This is a special case because we set scaling functions
// for both colors instead of only one.
e->scalingFunction[WHITE] = &ScaleKPKP[WHITE];
e->scalingFunction[BLACK] = &ScaleKPKP[BLACK];
}
}
// Zero or just one pawn makes it difficult to win, even with a small material
// advantage. This catches some trivial draws like KK, KBK and KNK and gives a
// drawish scale factor for cases such as KRKBP and KmmKm (except for KBBKN).
if (!pos.count<PAWN>(WHITE) && npm_w - npm_b <= BishopValueMg)
e->factor[WHITE] = uint8_t(npm_w < RookValueMg ? SCALE_FACTOR_DRAW :
npm_b <= BishopValueMg ? 4 : 14);
if (!pos.count<PAWN>(BLACK) && npm_b - npm_w <= BishopValueMg)
e->factor[BLACK] = uint8_t(npm_b < RookValueMg ? SCALE_FACTOR_DRAW :
npm_w <= BishopValueMg ? 4 : 14);
// Evaluate the material imbalance. We use PIECE_TYPE_NONE as a place holder
// for the bishop pair "extended piece", which allows us to be more flexible
// in defining bishop pair bonuses.
const int pieceCount[COLOR_NB][PIECE_TYPE_NB] = {
{ pos.count<BISHOP>(WHITE) > 1, pos.count<PAWN>(WHITE), pos.count<KNIGHT>(WHITE),
pos.count<BISHOP>(WHITE) , pos.count<ROOK>(WHITE), pos.count<QUEEN >(WHITE) },
{ pos.count<BISHOP>(BLACK) > 1, pos.count<PAWN>(BLACK), pos.count<KNIGHT>(BLACK),
pos.count<BISHOP>(BLACK) , pos.count<ROOK>(BLACK), pos.count<QUEEN >(BLACK) } };
e->score = (imbalance<WHITE>(pieceCount) - imbalance<BLACK>(pieceCount)) / 16;
return e;
}
} // namespace Material
} // namespace Stockfish

71
src/Stockfish/src/material.h Normal file
View file

@ -0,0 +1,71 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef MATERIAL_H_INCLUDED
#define MATERIAL_H_INCLUDED
#include "endgame.h"
#include "misc.h"
#include "position.h"
#include "types.h"
namespace Stockfish::Material {
/// Material::Entry contains various information about a material configuration.
/// It contains a material imbalance evaluation, a function pointer to a special
/// endgame evaluation function (which in most cases is nullptr, meaning that the
/// standard evaluation function will be used), and scale factors.
///
/// The scale factors are used to scale the evaluation score up or down. For
/// instance, in KRB vs KR endgames, the score is scaled down by a factor of 4,
/// which will result in scores of absolute value less than one pawn.
struct Entry {
Score imbalance() const { return score; }
Phase game_phase() const { return (Phase)gamePhase; }
bool specialized_eval_exists() const { return evaluationFunction != nullptr; }
Value evaluate(const Position& pos) const { return (*evaluationFunction)(pos); }
// scale_factor() takes a position and a color as input and returns a scale factor
// for the given color. We have to provide the position in addition to the color
// because the scale factor may also be a function which should be applied to
// the position. For instance, in KBP vs K endgames, the scaling function looks
// for rook pawns and wrong-colored bishops.
ScaleFactor scale_factor(const Position& pos, Color c) const {
ScaleFactor sf = scalingFunction[c] ? (*scalingFunction[c])(pos)
: SCALE_FACTOR_NONE;
return sf != SCALE_FACTOR_NONE ? sf : ScaleFactor(factor[c]);
}
Key key;
const EndgameBase<Value>* evaluationFunction;
const EndgameBase<ScaleFactor>* scalingFunction[COLOR_NB]; // Could be one for each
// side (e.g. KPKP, KBPsK)
Score score;
int16_t gamePhase;
uint8_t factor[COLOR_NB];
};
using Table = HashTable<Entry, 8192>;
Entry* probe(const Position& pos);
} // namespace Stockfish::Material
#endif // #ifndef MATERIAL_H_INCLUDED

787
src/Stockfish/src/misc.cpp Normal file
View file

@ -0,0 +1,787 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifdef _WIN32
#if _WIN32_WINNT < 0x0601
#undef _WIN32_WINNT
#define _WIN32_WINNT 0x0601 // Force to include needed API prototypes
#endif
#ifndef NOMINMAX
#define NOMINMAX
#endif
#include <windows.h>
// The needed Windows API for processor groups could be missed from old Windows
// versions, so instead of calling them directly (forcing the linker to resolve
// the calls at compile time), try to load them at runtime. To do this we need
// first to define the corresponding function pointers.
extern "C" {
using fun1_t = bool(*)(LOGICAL_PROCESSOR_RELATIONSHIP,
PSYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX, PDWORD);
using fun2_t = bool(*)(USHORT, PGROUP_AFFINITY);
using fun3_t = bool(*)(HANDLE, CONST GROUP_AFFINITY*, PGROUP_AFFINITY);
using fun4_t = bool(*)(USHORT, PGROUP_AFFINITY, USHORT, PUSHORT);
using fun5_t = WORD(*)();
using fun6_t = bool(*)(HANDLE, DWORD, PHANDLE);
using fun7_t = bool(*)(LPCSTR, LPCSTR, PLUID);
using fun8_t = bool(*)(HANDLE, BOOL, PTOKEN_PRIVILEGES, DWORD, PTOKEN_PRIVILEGES, PDWORD);
}
#endif
#include <cmath>
#include <cstdlib>
#include <fstream>
#include <iomanip>
#include <iostream>
#include <sstream>
#include <string_view>
#include <vector>
#if defined(__linux__) && !defined(__ANDROID__)
#include <stdlib.h>
#include <sys/mman.h>
#endif
#if defined(__APPLE__) || defined(__ANDROID__) || defined(__OpenBSD__) || (defined(__GLIBCXX__) && !defined(_GLIBCXX_HAVE_ALIGNED_ALLOC) && !defined(_WIN32)) || defined(__e2k__)
#define POSIXALIGNEDALLOC
#include <stdlib.h>
#endif
#include "misc.h"
#include "thread.h"
using namespace std;
namespace Stockfish {
namespace {
/// Version number or dev.
constexpr string_view version = "16";
/// Our fancy logging facility. The trick here is to replace cin.rdbuf() and
/// cout.rdbuf() with two Tie objects that tie cin and cout to a file stream. We
/// can toggle the logging of fakeout and std:cin at runtime whilst preserving
/// usual I/O functionality, all without changing a single line of code!
/// Idea from http://groups.google.com/group/comp.lang.c++/msg/1d941c0f26ea0d81
struct Tie: public streambuf { // MSVC requires split streambuf for cin and cout
Tie(streambuf* b, streambuf* l) : buf(b), logBuf(l) {}
int sync() override { return logBuf->pubsync(), buf->pubsync(); }
int overflow(int c) override { return log(buf->sputc((char)c), "<< "); }
int underflow() override { return buf->sgetc(); }
int uflow() override { return log(buf->sbumpc(), ">> "); }
streambuf *buf, *logBuf;
int log(int c, const char* prefix) {
static int last = '\n'; // Single log file
if (last == '\n')
logBuf->sputn(prefix, 3);
return last = logBuf->sputc((char)c);
}
};
class Logger {
Logger() : in(cin.rdbuf(), file.rdbuf()), out(cout.rdbuf(), file.rdbuf()) {}
~Logger() { start(""); }
ofstream file;
Tie in, out;
public:
static void start(const std::string& fname) {
static Logger l;
if (l.file.is_open())
{
cout.rdbuf(l.out.buf);
cin.rdbuf(l.in.buf);
l.file.close();
}
if (!fname.empty())
{
l.file.open(fname, ifstream::out);
if (!l.file.is_open())
{
cerr << "Unable to open debug log file " << fname << endl;
exit(EXIT_FAILURE);
}
cin.rdbuf(&l.in);
cout.rdbuf(&l.out);
}
}
};
} // namespace
/// engine_info() returns the full name of the current Stockfish version.
/// For local dev compiles we try to append the commit sha and commit date
/// from git if that fails only the local compilation date is set and "nogit" is specified:
/// Stockfish dev-YYYYMMDD-SHA
/// or
/// Stockfish dev-YYYYMMDD-nogit
///
/// For releases (non dev builds) we only include the version number:
/// Stockfish version
string engine_info(bool to_uci) {
stringstream ss;
ss << "Stockfish " << version << setfill('0');
if constexpr (version == "dev")
{
ss << "-";
#ifdef GIT_DATE
ss << stringify(GIT_DATE);
#else
constexpr string_view months("Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec");
string month, day, year;
stringstream date(__DATE__); // From compiler, format is "Sep 21 2008"
date >> month >> day >> year;
ss << year << setw(2) << setfill('0') << (1 + months.find(month) / 4) << setw(2) << setfill('0') << day;
#endif
ss << "-";
#ifdef GIT_SHA
ss << stringify(GIT_SHA);
#else
ss << "nogit";
#endif
}
ss << (to_uci ? "\nid author ": " by ")
<< "the Stockfish developers (see AUTHORS file)";
return ss.str();
}
/// compiler_info() returns a string trying to describe the compiler we use
std::string compiler_info() {
#define make_version_string(major, minor, patch) stringify(major) "." stringify(minor) "." stringify(patch)
/// Predefined macros hell:
///
/// __GNUC__ Compiler is gcc, Clang or Intel on Linux
/// __INTEL_COMPILER Compiler is Intel
/// _MSC_VER Compiler is MSVC or Intel on Windows
/// _WIN32 Building on Windows (any)
/// _WIN64 Building on Windows 64 bit
std::string compiler = "\nCompiled by ";
#ifdef __clang__
compiler += "clang++ ";
compiler += make_version_string(__clang_major__, __clang_minor__, __clang_patchlevel__);
#elif __INTEL_COMPILER
compiler += "Intel compiler ";
compiler += "(version ";
compiler += stringify(__INTEL_COMPILER) " update " stringify(__INTEL_COMPILER_UPDATE);
compiler += ")";
#elif _MSC_VER
compiler += "MSVC ";
compiler += "(version ";
compiler += stringify(_MSC_FULL_VER) "." stringify(_MSC_BUILD);
compiler += ")";
#elif defined(__e2k__) && defined(__LCC__)
#define dot_ver2(n) \
compiler += (char)'.'; \
compiler += (char)('0' + (n) / 10); \
compiler += (char)('0' + (n) % 10);
compiler += "MCST LCC ";
compiler += "(version ";
compiler += std::to_string(__LCC__ / 100);
dot_ver2(__LCC__ % 100)
dot_ver2(__LCC_MINOR__)
compiler += ")";
#elif __GNUC__
compiler += "g++ (GNUC) ";
compiler += make_version_string(__GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__);
#else
compiler += "Unknown compiler ";
compiler += "(unknown version)";
#endif
#if defined(__APPLE__)
compiler += " on Apple";
#elif defined(__CYGWIN__)
compiler += " on Cygwin";
#elif defined(__MINGW64__)
compiler += " on MinGW64";
#elif defined(__MINGW32__)
compiler += " on MinGW32";
#elif defined(__ANDROID__)
compiler += " on Android";
#elif defined(__linux__)
compiler += " on Linux";
#elif defined(_WIN64)
compiler += " on Microsoft Windows 64-bit";
#elif defined(_WIN32)
compiler += " on Microsoft Windows 32-bit";
#else
compiler += " on unknown system";
#endif
compiler += "\nCompilation settings include: ";
compiler += (Is64Bit ? " 64bit" : " 32bit");
#if defined(USE_VNNI)
compiler += " VNNI";
#endif
#if defined(USE_AVX512)
compiler += " AVX512";
#endif
compiler += (HasPext ? " BMI2" : "");
#if defined(USE_AVX2)
compiler += " AVX2";
#endif
#if defined(USE_SSE41)
compiler += " SSE41";
#endif
#if defined(USE_SSSE3)
compiler += " SSSE3";
#endif
#if defined(USE_SSE2)
compiler += " SSE2";
#endif
compiler += (HasPopCnt ? " POPCNT" : "");
#if defined(USE_MMX)
compiler += " MMX";
#endif
#if defined(USE_NEON)
compiler += " NEON";
#endif
#if !defined(NDEBUG)
compiler += " DEBUG";
#endif
compiler += "\n__VERSION__ macro expands to: ";
#ifdef __VERSION__
compiler += __VERSION__;
#else
compiler += "(undefined macro)";
#endif
compiler += "\n";
return compiler;
}
/// Debug functions used mainly to collect run-time statistics
constexpr int MaxDebugSlots = 32;
namespace {
template<size_t N>
struct DebugInfo {
std::atomic<int64_t> data[N] = { 0 };
constexpr inline std::atomic<int64_t>& operator[](int index) { return data[index]; }
};
DebugInfo<2> hit[MaxDebugSlots];
DebugInfo<2> mean[MaxDebugSlots];
DebugInfo<3> stdev[MaxDebugSlots];
DebugInfo<6> correl[MaxDebugSlots];
} // namespace
void dbg_hit_on(bool cond, int slot) {
++hit[slot][0];
if (cond)
++hit[slot][1];
}
void dbg_mean_of(int64_t value, int slot) {
++mean[slot][0];
mean[slot][1] += value;
}
void dbg_stdev_of(int64_t value, int slot) {
++stdev[slot][0];
stdev[slot][1] += value;
stdev[slot][2] += value * value;
}
void dbg_correl_of(int64_t value1, int64_t value2, int slot) {
++correl[slot][0];
correl[slot][1] += value1;
correl[slot][2] += value1 * value1;
correl[slot][3] += value2;
correl[slot][4] += value2 * value2;
correl[slot][5] += value1 * value2;
}
void dbg_print() {
int64_t n;
auto E = [&n](int64_t x) { return double(x) / n; };
auto sqr = [](double x) { return x * x; };
for (int i = 0; i < MaxDebugSlots; ++i)
if ((n = hit[i][0]))
std::cerr << "Hit #" << i
<< ": Total " << n << " Hits " << hit[i][1]
<< " Hit Rate (%) " << 100.0 * E(hit[i][1])
<< fakeendl;
for (int i = 0; i < MaxDebugSlots; ++i)
if ((n = mean[i][0]))
{
std::cerr << "Mean #" << i
<< ": Total " << n << " Mean " << E(mean[i][1])
<< fakeendl;
}
for (int i = 0; i < MaxDebugSlots; ++i)
if ((n = stdev[i][0]))
{
double r = sqrtl(E(stdev[i][2]) - sqr(E(stdev[i][1])));
std::cerr << "Stdev #" << i
<< ": Total " << n << " Stdev " << r
<< fakeendl;
}
for (int i = 0; i < MaxDebugSlots; ++i)
if ((n = correl[i][0]))
{
double r = (E(correl[i][5]) - E(correl[i][1]) * E(correl[i][3]))
/ ( sqrtl(E(correl[i][2]) - sqr(E(correl[i][1])))
* sqrtl(E(correl[i][4]) - sqr(E(correl[i][3]))));
std::cerr << "Correl. #" << i
<< ": Total " << n << " Coefficient " << r
<< fakeendl;
}
}
/// Used to serialize access to fakeout to avoid multiple threads writing at
/// the same time.
std::ostream& operator<<(std::ostream& os, SyncCout sc) {
static std::mutex m;
if (sc == IO_LOCK)
m.lock();
if (sc == IO_UNLOCK)
m.unlock();
return os;
}
/// Trampoline helper to avoid moving Logger to misc.h
void start_logger(const std::string& fname) { Logger::start(fname); }
/// prefetch() preloads the given address in L1/L2 cache. This is a non-blocking
/// function that doesn't stall the CPU waiting for data to be loaded from memory,
/// which can be quite slow.
#ifdef NO_PREFETCH
void prefetch(void*) {}
#else
void prefetch(void* addr) {
# if defined(__INTEL_COMPILER)
// This hack prevents prefetches from being optimized away by
// Intel compiler. Both MSVC and gcc seem not be affected by this.
__asm__ ("");
# endif
# if defined(__INTEL_COMPILER) || defined(_MSC_VER)
_mm_prefetch((char*)addr, _MM_HINT_T0);
# else
__builtin_prefetch(addr);
# endif
}
#endif
/// std_aligned_alloc() is our wrapper for systems where the c++17 implementation
/// does not guarantee the availability of aligned_alloc(). Memory allocated with
/// std_aligned_alloc() must be freed with std_aligned_free().
void* std_aligned_alloc(size_t alignment, size_t size) {
#if defined(POSIXALIGNEDALLOC)
void *mem;
return posix_memalign(&mem, alignment, size) ? nullptr : mem;
#elif defined(_WIN32) && !defined(_M_ARM) && !defined(_M_ARM64)
return _mm_malloc(size, alignment);
#elif defined(_WIN32)
return _aligned_malloc(size, alignment);
#else
return std::aligned_alloc(alignment, size);
#endif
}
void std_aligned_free(void* ptr) {
#if defined(POSIXALIGNEDALLOC)
free(ptr);
#elif defined(_WIN32) && !defined(_M_ARM) && !defined(_M_ARM64)
_mm_free(ptr);
#elif defined(_WIN32)
_aligned_free(ptr);
#else
free(ptr);
#endif
}
/// aligned_large_pages_alloc() will return suitably aligned memory, if possible using large pages.
#if defined(_WIN32)
static void* aligned_large_pages_alloc_windows([[maybe_unused]] size_t allocSize) {
#if !defined(_WIN64)
return nullptr;
#else
HANDLE hProcessToken { };
LUID luid { };
void* mem = nullptr;
const size_t largePageSize = GetLargePageMinimum();
if (!largePageSize)
return nullptr;
// Dynamically link OpenProcessToken, LookupPrivilegeValue and AdjustTokenPrivileges
HMODULE hAdvapi32 = GetModuleHandle(TEXT("advapi32.dll"));
if (!hAdvapi32)
hAdvapi32 = LoadLibrary(TEXT("advapi32.dll"));
auto fun6 = (fun6_t)(void(*)())GetProcAddress(hAdvapi32, "OpenProcessToken");
if (!fun6)
return nullptr;
auto fun7 = (fun7_t)(void(*)())GetProcAddress(hAdvapi32, "LookupPrivilegeValueA");
if (!fun7)
return nullptr;
auto fun8 = (fun8_t)(void(*)())GetProcAddress(hAdvapi32, "AdjustTokenPrivileges");
if (!fun8)
return nullptr;
// We need SeLockMemoryPrivilege, so try to enable it for the process
if (!fun6( // OpenProcessToken()
GetCurrentProcess(), TOKEN_ADJUST_PRIVILEGES | TOKEN_QUERY, &hProcessToken))
return nullptr;
if (fun7( // LookupPrivilegeValue(nullptr, SE_LOCK_MEMORY_NAME, &luid)
nullptr, "SeLockMemoryPrivilege", &luid))
{
TOKEN_PRIVILEGES tp { };
TOKEN_PRIVILEGES prevTp { };
DWORD prevTpLen = 0;
tp.PrivilegeCount = 1;
tp.Privileges[0].Luid = luid;
tp.Privileges[0].Attributes = SE_PRIVILEGE_ENABLED;
// Try to enable SeLockMemoryPrivilege. Note that even if AdjustTokenPrivileges() succeeds,
// we still need to query GetLastError() to ensure that the privileges were actually obtained.
if (fun8( // AdjustTokenPrivileges()
hProcessToken, FALSE, &tp, sizeof(TOKEN_PRIVILEGES), &prevTp, &prevTpLen) &&
GetLastError() == ERROR_SUCCESS)
{
// Round up size to full pages and allocate
allocSize = (allocSize + largePageSize - 1) & ~size_t(largePageSize - 1);
mem = VirtualAlloc(
nullptr, allocSize, MEM_RESERVE | MEM_COMMIT | MEM_LARGE_PAGES, PAGE_READWRITE);
// Privilege no longer needed, restore previous state
fun8( // AdjustTokenPrivileges ()
hProcessToken, FALSE, &prevTp, 0, nullptr, nullptr);
}
}
CloseHandle(hProcessToken);
return mem;
#endif
}
void* aligned_large_pages_alloc(size_t allocSize) {
// Try to allocate large pages
void* mem = aligned_large_pages_alloc_windows(allocSize);
// Fall back to regular, page aligned, allocation if necessary
if (!mem)
mem = VirtualAlloc(nullptr, allocSize, MEM_RESERVE | MEM_COMMIT, PAGE_READWRITE);
return mem;
}
#else
void* aligned_large_pages_alloc(size_t allocSize) {
#if defined(__linux__)
constexpr size_t alignment = 2 * 1024 * 1024; // assumed 2MB page size
#else
constexpr size_t alignment = 4096; // assumed small page size
#endif
// round up to multiples of alignment
size_t size = ((allocSize + alignment - 1) / alignment) * alignment;
void *mem = std_aligned_alloc(alignment, size);
#if defined(MADV_HUGEPAGE)
madvise(mem, size, MADV_HUGEPAGE);
#endif
return mem;
}
#endif
/// aligned_large_pages_free() will free the previously allocated ttmem
#if defined(_WIN32)
void aligned_large_pages_free(void* mem) {
if (mem && !VirtualFree(mem, 0, MEM_RELEASE))
{
DWORD err = GetLastError();
std::cerr << "Failed to free large page memory. Error code: 0x"
<< std::hex << err
<< std::dec << fakeendl;
exit(EXIT_FAILURE);
}
}
#else
void aligned_large_pages_free(void *mem) {
std_aligned_free(mem);
}
#endif
namespace WinProcGroup {
#ifndef _WIN32
void bindThisThread(size_t) {}
#else
/// best_node() retrieves logical processor information using Windows specific
/// API and returns the best node id for the thread with index idx. Original
/// code from Texel by Peter Österlund.
static int best_node(size_t idx) {
int threads = 0;
int nodes = 0;
int cores = 0;
DWORD returnLength = 0;
DWORD byteOffset = 0;
// Early exit if the needed API is not available at runtime
HMODULE k32 = GetModuleHandle(TEXT("Kernel32.dll"));
auto fun1 = (fun1_t)(void(*)())GetProcAddress(k32, "GetLogicalProcessorInformationEx");
if (!fun1)
return -1;
// First call to GetLogicalProcessorInformationEx() to get returnLength.
// We expect the call to fail due to null buffer.
if (fun1(RelationAll, nullptr, &returnLength))
return -1;
// Once we know returnLength, allocate the buffer
SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX *buffer, *ptr;
ptr = buffer = (SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*)malloc(returnLength);
// Second call to GetLogicalProcessorInformationEx(), now we expect to succeed
if (!fun1(RelationAll, buffer, &returnLength))
{
free(buffer);
return -1;
}
while (byteOffset < returnLength)
{
if (ptr->Relationship == RelationNumaNode)
nodes++;
else if (ptr->Relationship == RelationProcessorCore)
{
cores++;
threads += (ptr->Processor.Flags == LTP_PC_SMT) ? 2 : 1;
}
assert(ptr->Size);
byteOffset += ptr->Size;
ptr = (SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX*)(((char*)ptr) + ptr->Size);
}
free(buffer);
std::vector<int> groups;
// Run as many threads as possible on the same node until core limit is
// reached, then move on filling the next node.
for (int n = 0; n < nodes; n++)
for (int i = 0; i < cores / nodes; i++)
groups.push_back(n);
// In case a core has more than one logical processor (we assume 2) and we
// have still threads to allocate, then spread them evenly across available
// nodes.
for (int t = 0; t < threads - cores; t++)
groups.push_back(t % nodes);
// If we still have more threads than the total number of logical processors
// then return -1 and let the OS to decide what to do.
return idx < groups.size() ? groups[idx] : -1;
}
/// bindThisThread() set the group affinity of the current thread
void bindThisThread(size_t idx) {
// Use only local variables to be thread-safe
int node = best_node(idx);
if (node == -1)
return;
// Early exit if the needed API are not available at runtime
HMODULE k32 = GetModuleHandle(TEXT("Kernel32.dll"));
auto fun2 = (fun2_t)(void(*)())GetProcAddress(k32, "GetNumaNodeProcessorMaskEx");
auto fun3 = (fun3_t)(void(*)())GetProcAddress(k32, "SetThreadGroupAffinity");
auto fun4 = (fun4_t)(void(*)())GetProcAddress(k32, "GetNumaNodeProcessorMask2");
auto fun5 = (fun5_t)(void(*)())GetProcAddress(k32, "GetMaximumProcessorGroupCount");
if (!fun2 || !fun3)
return;
if (!fun4 || !fun5)
{
GROUP_AFFINITY affinity;
if (fun2(node, &affinity)) // GetNumaNodeProcessorMaskEx
fun3(GetCurrentThread(), &affinity, nullptr); // SetThreadGroupAffinity
}
else
{
// If a numa node has more than one processor group, we assume they are
// sized equal and we spread threads evenly across the groups.
USHORT elements, returnedElements;
elements = fun5(); // GetMaximumProcessorGroupCount
GROUP_AFFINITY *affinity = (GROUP_AFFINITY*)malloc(elements * sizeof(GROUP_AFFINITY));
if (fun4(node, affinity, elements, &returnedElements)) // GetNumaNodeProcessorMask2
fun3(GetCurrentThread(), &affinity[idx % returnedElements], nullptr); // SetThreadGroupAffinity
free(affinity);
}
}
#endif
} // namespace WinProcGroup
#ifdef _WIN32
#include <direct.h>
#define GETCWD _getcwd
#else
#include <unistd.h>
#define GETCWD getcwd
#endif
namespace CommandLine {
string argv0; // path+name of the executable binary, as given by argv[0]
string binaryDirectory; // path of the executable directory
string workingDirectory; // path of the working directory
void init([[maybe_unused]] int argc, char* argv[]) {
string pathSeparator;
// extract the path+name of the executable binary
argv0 = argv[0];
#ifdef _WIN32
pathSeparator = "\\";
#ifdef _MSC_VER
// Under windows argv[0] may not have the extension. Also _get_pgmptr() had
// issues in some windows 10 versions, so check returned values carefully.
char* pgmptr = nullptr;
if (!_get_pgmptr(&pgmptr) && pgmptr != nullptr && *pgmptr)
argv0 = pgmptr;
#endif
#else
pathSeparator = "/";
#endif
// extract the working directory
workingDirectory = "";
char buff[40000];
char* cwd = GETCWD(buff, 40000);
if (cwd)
workingDirectory = cwd;
// extract the binary directory path from argv0
binaryDirectory = argv0;
size_t pos = binaryDirectory.find_last_of("\\/");
if (pos == std::string::npos)
binaryDirectory = "." + pathSeparator;
else
binaryDirectory.resize(pos + 1);
// pattern replacement: "./" at the start of path is replaced by the working directory
if (binaryDirectory.find("." + pathSeparator) == 0)
binaryDirectory.replace(0, 1, workingDirectory);
}
} // namespace CommandLine
} // namespace Stockfish

176
src/Stockfish/src/misc.h Normal file
View file

@ -0,0 +1,176 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef MISC_H_INCLUDED
#define MISC_H_INCLUDED
#include <cassert>
#include <chrono>
#include <ostream>
#include <string>
#include <vector>
#include <cstdint>
#include "types.h"
#define stringify2(x) #x
#define stringify(x) stringify2(x)
namespace Stockfish {
std::string engine_info(bool to_uci = false);
std::string compiler_info();
void prefetch(void* addr);
void start_logger(const std::string& fname);
void* std_aligned_alloc(size_t alignment, size_t size);
void std_aligned_free(void* ptr);
void* aligned_large_pages_alloc(size_t size); // memory aligned by page size, min alignment: 4096 bytes
void aligned_large_pages_free(void* mem); // nop if mem == nullptr
void dbg_hit_on(bool cond, int slot = 0);
void dbg_mean_of(int64_t value, int slot = 0);
void dbg_stdev_of(int64_t value, int slot = 0);
void dbg_correl_of(int64_t value1, int64_t value2, int slot = 0);
void dbg_print();
using TimePoint = std::chrono::milliseconds::rep; // A value in milliseconds
static_assert(sizeof(TimePoint) == sizeof(int64_t), "TimePoint should be 64 bits");
inline TimePoint now() {
return std::chrono::duration_cast<std::chrono::milliseconds>
(std::chrono::steady_clock::now().time_since_epoch()).count();
}
template<class Entry, int Size>
struct HashTable {
Entry* operator[](Key key) { return &table[(uint32_t)key & (Size - 1)]; }
private:
std::vector<Entry> table = std::vector<Entry>(Size); // Allocate on the heap
};
enum SyncCout { IO_LOCK, IO_UNLOCK };
std::ostream& operator<<(std::ostream&, SyncCout);
#define sync_cout fakeout << IO_LOCK
#define sync_endl fakeendl << IO_UNLOCK
// align_ptr_up() : get the first aligned element of an array.
// ptr must point to an array of size at least `sizeof(T) * N + alignment` bytes,
// where N is the number of elements in the array.
template <uintptr_t Alignment, typename T>
T* align_ptr_up(T* ptr)
{
static_assert(alignof(T) < Alignment);
const uintptr_t ptrint = reinterpret_cast<uintptr_t>(reinterpret_cast<char*>(ptr));
return reinterpret_cast<T*>(reinterpret_cast<char*>((ptrint + (Alignment - 1)) / Alignment * Alignment));
}
// IsLittleEndian : true if and only if the binary is compiled on a little endian machine
static inline const union { uint32_t i; char c[4]; } Le = { 0x01020304 };
static inline const bool IsLittleEndian = (Le.c[0] == 4);
template <typename T, std::size_t MaxSize>
class ValueList {
public:
std::size_t size() const { return size_; }
void push_back(const T& value) { values_[size_++] = value; }
const T* begin() const { return values_; }
const T* end() const { return values_ + size_; }
private:
T values_[MaxSize];
std::size_t size_ = 0;
};
/// xorshift64star Pseudo-Random Number Generator
/// This class is based on original code written and dedicated
/// to the public domain by Sebastiano Vigna (2014).
/// It has the following characteristics:
///
/// - Outputs 64-bit numbers
/// - Passes Dieharder and SmallCrush test batteries
/// - Does not require warm-up, no zeroland to escape
/// - Internal state is a single 64-bit integer
/// - Period is 2^64 - 1
/// - Speed: 1.60 ns/call (Core i7 @3.40GHz)
///
/// For further analysis see
/// <http://vigna.di.unimi.it/ftp/papers/xorshift.pdf>
class PRNG {
uint64_t s;
uint64_t rand64() {
s ^= s >> 12, s ^= s << 25, s ^= s >> 27;
return s * 2685821657736338717LL;
}
public:
PRNG(uint64_t seed) : s(seed) { assert(seed); }
template<typename T> T rand() { return T(rand64()); }
/// Special generator used to fast init magic numbers.
/// Output values only have 1/8th of their bits set on average.
template<typename T> T sparse_rand()
{ return T(rand64() & rand64() & rand64()); }
};
inline uint64_t mul_hi64(uint64_t a, uint64_t b) {
#if defined(__GNUC__) && defined(IS_64BIT)
__extension__ using uint128 = unsigned __int128;
return ((uint128)a * (uint128)b) >> 64;
#else
uint64_t aL = (uint32_t)a, aH = a >> 32;
uint64_t bL = (uint32_t)b, bH = b >> 32;
uint64_t c1 = (aL * bL) >> 32;
uint64_t c2 = aH * bL + c1;
uint64_t c3 = aL * bH + (uint32_t)c2;
return aH * bH + (c2 >> 32) + (c3 >> 32);
#endif
}
/// Under Windows it is not possible for a process to run on more than one
/// logical processor group. This usually means to be limited to use max 64
/// cores. To overcome this, some special platform specific API should be
/// called to set group affinity for each thread. Original code from Texel by
/// Peter Österlund.
namespace WinProcGroup {
void bindThisThread(size_t idx);
}
namespace CommandLine {
void init(int argc, char* argv[]);
extern std::string binaryDirectory; // path of the executable directory
extern std::string workingDirectory; // path of the working directory
}
} // namespace Stockfish
#endif // #ifndef MISC_H_INCLUDED

284
src/Stockfish/src/movegen.cpp Normal file
View file

@ -0,0 +1,284 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cassert>
#include "movegen.h"
#include "position.h"
namespace Stockfish {
namespace {
template<GenType Type, Direction D, bool Enemy>
ExtMove* make_promotions(ExtMove* moveList, [[maybe_unused]] Square to) {
if constexpr (Type == CAPTURES || Type == EVASIONS || Type == NON_EVASIONS)
{
*moveList++ = make<PROMOTION>(to - D, to, QUEEN);
if constexpr (Enemy && Type == CAPTURES)
{
*moveList++ = make<PROMOTION>(to - D, to, ROOK);
*moveList++ = make<PROMOTION>(to - D, to, BISHOP);
*moveList++ = make<PROMOTION>(to - D, to, KNIGHT);
}
}
if constexpr ((Type == QUIETS && !Enemy) || Type == EVASIONS || Type == NON_EVASIONS)
{
*moveList++ = make<PROMOTION>(to - D, to, ROOK);
*moveList++ = make<PROMOTION>(to - D, to, BISHOP);
*moveList++ = make<PROMOTION>(to - D, to, KNIGHT);
}
return moveList;
}
template<Color Us, GenType Type>
ExtMove* generate_pawn_moves(const Position& pos, ExtMove* moveList, Bitboard target) {
constexpr Color Them = ~Us;
constexpr Bitboard TRank7BB = (Us == WHITE ? Rank7BB : Rank2BB);
constexpr Bitboard TRank3BB = (Us == WHITE ? Rank3BB : Rank6BB);
constexpr Direction Up = pawn_push(Us);
constexpr Direction UpRight = (Us == WHITE ? NORTH_EAST : SOUTH_WEST);
constexpr Direction UpLeft = (Us == WHITE ? NORTH_WEST : SOUTH_EAST);
const Bitboard emptySquares = ~pos.pieces();
const Bitboard enemies = Type == EVASIONS ? pos.checkers()
: pos.pieces(Them);
Bitboard pawnsOn7 = pos.pieces(Us, PAWN) & TRank7BB;
Bitboard pawnsNotOn7 = pos.pieces(Us, PAWN) & ~TRank7BB;
// Single and double pawn pushes, no promotions
if constexpr (Type != CAPTURES)
{
Bitboard b1 = shift<Up>(pawnsNotOn7) & emptySquares;
Bitboard b2 = shift<Up>(b1 & TRank3BB) & emptySquares;
if constexpr (Type == EVASIONS) // Consider only blocking squares
{
b1 &= target;
b2 &= target;
}
if constexpr (Type == QUIET_CHECKS)
{
// To make a quiet check, you either make a direct check by pushing a pawn
// or push a blocker pawn that is not on the same file as the enemy king.
// Discovered check promotion has been already generated amongst the captures.
Square ksq = pos.square<KING>(Them);
Bitboard dcCandidatePawns = pos.blockers_for_king(Them) & ~file_bb(ksq);
b1 &= pawn_attacks_bb(Them, ksq) | shift< Up>(dcCandidatePawns);
b2 &= pawn_attacks_bb(Them, ksq) | shift<Up+Up>(dcCandidatePawns);
}
while (b1)
{
Square to = pop_lsb(b1);
*moveList++ = make_move(to - Up, to);
}
while (b2)
{
Square to = pop_lsb(b2);
*moveList++ = make_move(to - Up - Up, to);
}
}
// Promotions and underpromotions
if (pawnsOn7)
{
Bitboard b1 = shift<UpRight>(pawnsOn7) & enemies;
Bitboard b2 = shift<UpLeft >(pawnsOn7) & enemies;
Bitboard b3 = shift<Up >(pawnsOn7) & emptySquares;
if constexpr (Type == EVASIONS)
b3 &= target;
while (b1)
moveList = make_promotions<Type, UpRight, true>(moveList, pop_lsb(b1));
while (b2)
moveList = make_promotions<Type, UpLeft, true>(moveList, pop_lsb(b2));
while (b3)
moveList = make_promotions<Type, Up, false>(moveList, pop_lsb(b3));
}
// Standard and en passant captures
if constexpr (Type == CAPTURES || Type == EVASIONS || Type == NON_EVASIONS)
{
Bitboard b1 = shift<UpRight>(pawnsNotOn7) & enemies;
Bitboard b2 = shift<UpLeft >(pawnsNotOn7) & enemies;
while (b1)
{
Square to = pop_lsb(b1);
*moveList++ = make_move(to - UpRight, to);
}
while (b2)
{
Square to = pop_lsb(b2);
*moveList++ = make_move(to - UpLeft, to);
}
if (pos.ep_square() != SQ_NONE)
{
assert(rank_of(pos.ep_square()) == relative_rank(Us, RANK_6));
// An en passant capture cannot resolve a discovered check
if (Type == EVASIONS && (target & (pos.ep_square() + Up)))
return moveList;
b1 = pawnsNotOn7 & pawn_attacks_bb(Them, pos.ep_square());
assert(b1);
while (b1)
*moveList++ = make<EN_PASSANT>(pop_lsb(b1), pos.ep_square());
}
}
return moveList;
}
template<Color Us, PieceType Pt, bool Checks>
ExtMove* generate_moves(const Position& pos, ExtMove* moveList, Bitboard target) {
static_assert(Pt != KING && Pt != PAWN, "Unsupported piece type in generate_moves()");
Bitboard bb = pos.pieces(Us, Pt);
while (bb)
{
Square from = pop_lsb(bb);
Bitboard b = attacks_bb<Pt>(from, pos.pieces()) & target;
// To check, you either move freely a blocker or make a direct check.
if (Checks && (Pt == QUEEN || !(pos.blockers_for_king(~Us) & from)))
b &= pos.check_squares(Pt);
while (b)
*moveList++ = make_move(from, pop_lsb(b));
}
return moveList;
}
template<Color Us, GenType Type>
ExtMove* generate_all(const Position& pos, ExtMove* moveList) {
static_assert(Type != LEGAL, "Unsupported type in generate_all()");
constexpr bool Checks = Type == QUIET_CHECKS; // Reduce template instantiations
const Square ksq = pos.square<KING>(Us);
Bitboard target;
// Skip generating non-king moves when in double check
if (Type != EVASIONS || !more_than_one(pos.checkers()))
{
target = Type == EVASIONS ? between_bb(ksq, lsb(pos.checkers()))
: Type == NON_EVASIONS ? ~pos.pieces( Us)
: Type == CAPTURES ? pos.pieces(~Us)
: ~pos.pieces( ); // QUIETS || QUIET_CHECKS
moveList = generate_pawn_moves<Us, Type>(pos, moveList, target);
moveList = generate_moves<Us, KNIGHT, Checks>(pos, moveList, target);
moveList = generate_moves<Us, BISHOP, Checks>(pos, moveList, target);
moveList = generate_moves<Us, ROOK, Checks>(pos, moveList, target);
moveList = generate_moves<Us, QUEEN, Checks>(pos, moveList, target);
}
if (!Checks || pos.blockers_for_king(~Us) & ksq)
{
Bitboard b = attacks_bb<KING>(ksq) & (Type == EVASIONS ? ~pos.pieces(Us) : target);
if (Checks)
b &= ~attacks_bb<QUEEN>(pos.square<KING>(~Us));
while (b)
*moveList++ = make_move(ksq, pop_lsb(b));
if ((Type == QUIETS || Type == NON_EVASIONS) && pos.can_castle(Us & ANY_CASTLING))
for (CastlingRights cr : { Us & KING_SIDE, Us & QUEEN_SIDE } )
if (!pos.castling_impeded(cr) && pos.can_castle(cr))
*moveList++ = make<CASTLING>(ksq, pos.castling_rook_square(cr));
}
return moveList;
}
} // namespace
/// <CAPTURES> Generates all pseudo-legal captures plus queen promotions
/// <QUIETS> Generates all pseudo-legal non-captures and underpromotions
/// <EVASIONS> Generates all pseudo-legal check evasions when the side to move is in check
/// <QUIET_CHECKS> Generates all pseudo-legal non-captures giving check, except castling and promotions
/// <NON_EVASIONS> Generates all pseudo-legal captures and non-captures
///
/// Returns a pointer to the end of the move list.
template<GenType Type>
ExtMove* generate(const Position& pos, ExtMove* moveList) {
static_assert(Type != LEGAL, "Unsupported type in generate()");
assert((Type == EVASIONS) == (bool)pos.checkers());
Color us = pos.side_to_move();
return us == WHITE ? generate_all<WHITE, Type>(pos, moveList)
: generate_all<BLACK, Type>(pos, moveList);
}
// Explicit template instantiations
template ExtMove* generate<CAPTURES>(const Position&, ExtMove*);
template ExtMove* generate<QUIETS>(const Position&, ExtMove*);
template ExtMove* generate<EVASIONS>(const Position&, ExtMove*);
template ExtMove* generate<QUIET_CHECKS>(const Position&, ExtMove*);
template ExtMove* generate<NON_EVASIONS>(const Position&, ExtMove*);
/// generate<LEGAL> generates all the legal moves in the given position
template<>
ExtMove* generate<LEGAL>(const Position& pos, ExtMove* moveList) {
Color us = pos.side_to_move();
Bitboard pinned = pos.blockers_for_king(us) & pos.pieces(us);
Square ksq = pos.square<KING>(us);
ExtMove* cur = moveList;
moveList = pos.checkers() ? generate<EVASIONS >(pos, moveList)
: generate<NON_EVASIONS>(pos, moveList);
while (cur != moveList)
if ( ((pinned & from_sq(*cur)) || from_sq(*cur) == ksq || type_of(*cur) == EN_PASSANT)
&& !pos.legal(*cur))
*cur = (--moveList)->move;
else
++cur;
return moveList;
}
} // namespace Stockfish

77
src/Stockfish/src/movegen.h Normal file
View file

@ -0,0 +1,77 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef MOVEGEN_H_INCLUDED
#define MOVEGEN_H_INCLUDED
#include <algorithm>
#include "types.h"
namespace Stockfish {
class Position;
enum GenType {
CAPTURES,
QUIETS,
QUIET_CHECKS,
EVASIONS,
NON_EVASIONS,
LEGAL
};
struct ExtMove {
Move move;
int value;
operator Move() const { return move; }
void operator=(Move m) { move = m; }
// Inhibit unwanted implicit conversions to Move
// with an ambiguity that yields to a compile error.
operator float() const = delete;
};
inline bool operator<(const ExtMove& f, const ExtMove& s) {
return f.value < s.value;
}
template<GenType>
ExtMove* generate(const Position& pos, ExtMove* moveList);
/// The MoveList struct is a simple wrapper around generate(). It sometimes comes
/// in handy to use this class instead of the low level generate() function.
template<GenType T>
struct MoveList {
explicit MoveList(const Position& pos) : last(generate<T>(pos, moveList)) {}
const ExtMove* begin() const { return moveList; }
const ExtMove* end() const { return last; }
size_t size() const { return last - moveList; }
bool contains(Move move) const {
return std::find(begin(), end(), move) != end();
}
private:
ExtMove moveList[MAX_MOVES], *last;
};
} // namespace Stockfish
#endif // #ifndef MOVEGEN_H_INCLUDED

295
src/Stockfish/src/movepick.cpp Normal file
View file

@ -0,0 +1,295 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cassert>
#include "bitboard.h"
#include "movepick.h"
namespace Stockfish {
namespace {
enum Stages {
MAIN_TT, CAPTURE_INIT, GOOD_CAPTURE, REFUTATION, QUIET_INIT, QUIET, BAD_CAPTURE,
EVASION_TT, EVASION_INIT, EVASION,
PROBCUT_TT, PROBCUT_INIT, PROBCUT,
QSEARCH_TT, QCAPTURE_INIT, QCAPTURE, QCHECK_INIT, QCHECK
};
// partial_insertion_sort() sorts moves in descending order up to and including
// a given limit. The order of moves smaller than the limit is left unspecified.
void partial_insertion_sort(ExtMove* begin, ExtMove* end, int limit) {
for (ExtMove *sortedEnd = begin, *p = begin + 1; p < end; ++p)
if (p->value >= limit)
{
ExtMove tmp = *p, *q;
*p = *++sortedEnd;
for (q = sortedEnd; q != begin && *(q - 1) < tmp; --q)
*q = *(q - 1);
*q = tmp;
}
}
} // namespace
/// Constructors of the MovePicker class. As arguments we pass information
/// to help it to return the (presumably) good moves first, to decide which
/// moves to return (in the quiescence search, for instance, we only want to
/// search captures, promotions, and some checks) and how important good move
/// ordering is at the current node.
/// MovePicker constructor for the main search
MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const ButterflyHistory* mh,
const CapturePieceToHistory* cph,
const PieceToHistory** ch,
Move cm,
const Move* killers)
: pos(p), mainHistory(mh), captureHistory(cph), continuationHistory(ch),
ttMove(ttm), refutations{{killers[0], 0}, {killers[1], 0}, {cm, 0}}, depth(d)
{
assert(d > 0);
stage = (pos.checkers() ? EVASION_TT : MAIN_TT) +
!(ttm && pos.pseudo_legal(ttm));
}
/// MovePicker constructor for quiescence search
MovePicker::MovePicker(const Position& p, Move ttm, Depth d, const ButterflyHistory* mh,
const CapturePieceToHistory* cph,
const PieceToHistory** ch,
Square rs)
: pos(p), mainHistory(mh), captureHistory(cph), continuationHistory(ch), ttMove(ttm), recaptureSquare(rs), depth(d)
{
assert(d <= 0);
stage = (pos.checkers() ? EVASION_TT : QSEARCH_TT) +
!( ttm
&& pos.pseudo_legal(ttm));
}
/// MovePicker constructor for ProbCut: we generate captures with SEE greater
/// than or equal to the given threshold.
MovePicker::MovePicker(const Position& p, Move ttm, Value th, const CapturePieceToHistory* cph)
: pos(p), captureHistory(cph), ttMove(ttm), threshold(th)
{
assert(!pos.checkers());
stage = PROBCUT_TT + !(ttm && pos.capture_stage(ttm)
&& pos.pseudo_legal(ttm)
&& pos.see_ge(ttm, threshold));
}
/// MovePicker::score() assigns a numerical value to each move in a list, used
/// for sorting. Captures are ordered by Most Valuable Victim (MVV), preferring
/// captures with a good history. Quiets moves are ordered using the history tables.
template<GenType Type>
void MovePicker::score() {
static_assert(Type == CAPTURES || Type == QUIETS || Type == EVASIONS, "Wrong type");
[[maybe_unused]] Bitboard threatenedByPawn, threatenedByMinor, threatenedByRook, threatenedPieces;
if constexpr (Type == QUIETS)
{
Color us = pos.side_to_move();
threatenedByPawn = pos.attacks_by<PAWN>(~us);
threatenedByMinor = pos.attacks_by<KNIGHT>(~us) | pos.attacks_by<BISHOP>(~us) | threatenedByPawn;
threatenedByRook = pos.attacks_by<ROOK>(~us) | threatenedByMinor;
// Pieces threatened by pieces of lesser material value
threatenedPieces = (pos.pieces(us, QUEEN) & threatenedByRook)
| (pos.pieces(us, ROOK) & threatenedByMinor)
| (pos.pieces(us, KNIGHT, BISHOP) & threatenedByPawn);
}
for (auto& m : *this)
if constexpr (Type == CAPTURES)
m.value = (7 * int(PieceValue[MG][pos.piece_on(to_sq(m))])
+ (*captureHistory)[pos.moved_piece(m)][to_sq(m)][type_of(pos.piece_on(to_sq(m)))]) / 16;
else if constexpr (Type == QUIETS)
m.value = 2 * (*mainHistory)[pos.side_to_move()][from_to(m)]
+ 2 * (*continuationHistory[0])[pos.moved_piece(m)][to_sq(m)]
+ (*continuationHistory[1])[pos.moved_piece(m)][to_sq(m)]
+ (*continuationHistory[3])[pos.moved_piece(m)][to_sq(m)]
+ (*continuationHistory[5])[pos.moved_piece(m)][to_sq(m)]
+ (threatenedPieces & from_sq(m) ?
(type_of(pos.moved_piece(m)) == QUEEN && !(to_sq(m) & threatenedByRook) ? 50000
: type_of(pos.moved_piece(m)) == ROOK && !(to_sq(m) & threatenedByMinor) ? 25000
: !(to_sq(m) & threatenedByPawn) ? 15000
: 0)
: 0)
+ bool(pos.check_squares(type_of(pos.moved_piece(m))) & to_sq(m)) * 16384;
else // Type == EVASIONS
{
if (pos.capture_stage(m))
m.value = PieceValue[MG][pos.piece_on(to_sq(m))]
- Value(type_of(pos.moved_piece(m)))
+ (1 << 28);
else
m.value = (*mainHistory)[pos.side_to_move()][from_to(m)]
+ (*continuationHistory[0])[pos.moved_piece(m)][to_sq(m)];
}
}
/// MovePicker::select() returns the next move satisfying a predicate function.
/// It never returns the TT move.
template<MovePicker::PickType T, typename Pred>
Move MovePicker::select(Pred filter) {
while (cur < endMoves)
{
if constexpr (T == Best)
std::swap(*cur, *std::max_element(cur, endMoves));
if (*cur != ttMove && filter())
return *cur++;
cur++;
}
return MOVE_NONE;
}
/// MovePicker::next_move() is the most important method of the MovePicker class. It
/// returns a new pseudo-legal move every time it is called until there are no more
/// moves left, picking the move with the highest score from a list of generated moves.
Move MovePicker::next_move(bool skipQuiets) {
top:
switch (stage) {
case MAIN_TT:
case EVASION_TT:
case QSEARCH_TT:
case PROBCUT_TT:
++stage;
return ttMove;
case CAPTURE_INIT:
case PROBCUT_INIT:
case QCAPTURE_INIT:
cur = endBadCaptures = moves;
endMoves = generate<CAPTURES>(pos, cur);
score<CAPTURES>();
partial_insertion_sort(cur, endMoves, std::numeric_limits<int>::min());
++stage;
goto top;
case GOOD_CAPTURE:
if (select<Next>([&](){
return pos.see_ge(*cur, Value(-cur->value)) ?
// Move losing capture to endBadCaptures to be tried later
true : (*endBadCaptures++ = *cur, false); }))
return *(cur - 1);
// Prepare the pointers to loop over the refutations array
cur = std::begin(refutations);
endMoves = std::end(refutations);
// If the countermove is the same as a killer, skip it
if ( refutations[0].move == refutations[2].move
|| refutations[1].move == refutations[2].move)
--endMoves;
++stage;
[[fallthrough]];
case REFUTATION:
if (select<Next>([&](){ return *cur != MOVE_NONE
&& !pos.capture_stage(*cur)
&& pos.pseudo_legal(*cur); }))
return *(cur - 1);
++stage;
[[fallthrough]];
case QUIET_INIT:
if (!skipQuiets)
{
cur = endBadCaptures;
endMoves = generate<QUIETS>(pos, cur);
score<QUIETS>();
partial_insertion_sort(cur, endMoves, -3000 * depth);
}
++stage;
[[fallthrough]];
case QUIET:
if ( !skipQuiets
&& select<Next>([&](){return *cur != refutations[0].move
&& *cur != refutations[1].move
&& *cur != refutations[2].move;}))
return *(cur - 1);
// Prepare the pointers to loop over the bad captures
cur = moves;
endMoves = endBadCaptures;
++stage;
[[fallthrough]];
case BAD_CAPTURE:
return select<Next>([](){ return true; });
case EVASION_INIT:
cur = moves;
endMoves = generate<EVASIONS>(pos, cur);
score<EVASIONS>();
++stage;
[[fallthrough]];
case EVASION:
return select<Best>([](){ return true; });
case PROBCUT:
return select<Next>([&](){ return pos.see_ge(*cur, threshold); });
case QCAPTURE:
if (select<Next>([&](){ return depth > DEPTH_QS_RECAPTURES
|| to_sq(*cur) == recaptureSquare; }))
return *(cur - 1);
// If we did not find any move and we do not try checks, we have finished
if (depth != DEPTH_QS_CHECKS)
return MOVE_NONE;
++stage;
[[fallthrough]];
case QCHECK_INIT:
cur = moves;
endMoves = generate<QUIET_CHECKS>(pos, cur);
++stage;
[[fallthrough]];
case QCHECK:
return select<Next>([](){ return true; });
}
assert(false);
return MOVE_NONE; // Silence warning
}
} // namespace Stockfish

155
src/Stockfish/src/movepick.h Normal file
View file

@ -0,0 +1,155 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef MOVEPICK_H_INCLUDED
#define MOVEPICK_H_INCLUDED
#include <array>
#include <limits>
#include <type_traits>
#include "movegen.h"
#include "position.h"
#include "types.h"
namespace Stockfish {
/// StatsEntry stores the stat table value. It is usually a number but could
/// be a move or even a nested history. We use a class instead of naked value
/// to directly call history update operator<<() on the entry so to use stats
/// tables at caller sites as simple multi-dim arrays.
template<typename T, int D>
class StatsEntry {
T entry;
public:
void operator=(const T& v) { entry = v; }
T* operator&() { return &entry; }
T* operator->() { return &entry; }
operator const T&() const { return entry; }
void operator<<(int bonus) {
assert(abs(bonus) <= D); // Ensure range is [-D, D]
static_assert(D <= std::numeric_limits<T>::max(), "D overflows T");
entry += bonus - entry * abs(bonus) / D;
assert(abs(entry) <= D);
}
};
/// Stats is a generic N-dimensional array used to store various statistics.
/// The first template parameter T is the base type of the array, the second
/// template parameter D limits the range of updates in [-D, D] when we update
/// values with the << operator, while the last parameters (Size and Sizes)
/// encode the dimensions of the array.
template <typename T, int D, int Size, int... Sizes>
struct Stats : public std::array<Stats<T, D, Sizes...>, Size>
{
using stats = Stats<T, D, Size, Sizes...>;
void fill(const T& v) {
// For standard-layout 'this' points to first struct member
assert(std::is_standard_layout<stats>::value);
using entry = StatsEntry<T, D>;
entry* p = reinterpret_cast<entry*>(this);
std::fill(p, p + sizeof(*this) / sizeof(entry), v);
}
};
template <typename T, int D, int Size>
struct Stats<T, D, Size> : public std::array<StatsEntry<T, D>, Size> {};
/// In stats table, D=0 means that the template parameter is not used
enum StatsParams { NOT_USED = 0 };
enum StatsType { NoCaptures, Captures };
/// ButterflyHistory records how often quiet moves have been successful or
/// unsuccessful during the current search, and is used for reduction and move
/// ordering decisions. It uses 2 tables (one for each color) indexed by
/// the move's from and to squares, see www.chessprogramming.org/Butterfly_Boards
/// (~11 elo)
using ButterflyHistory = Stats<int16_t, 7183, COLOR_NB, int(SQUARE_NB) * int(SQUARE_NB)>;
/// CounterMoveHistory stores counter moves indexed by [piece][to] of the previous
/// move, see www.chessprogramming.org/Countermove_Heuristic
using CounterMoveHistory = Stats<Move, NOT_USED, PIECE_NB, SQUARE_NB>;
/// CapturePieceToHistory is addressed by a move's [piece][to][captured piece type]
using CapturePieceToHistory = Stats<int16_t, 10692, PIECE_NB, SQUARE_NB, PIECE_TYPE_NB>;
/// PieceToHistory is like ButterflyHistory but is addressed by a move's [piece][to]
using PieceToHistory = Stats<int16_t, 29952, PIECE_NB, SQUARE_NB>;
/// ContinuationHistory is the combined history of a given pair of moves, usually
/// the current one given a previous one. The nested history table is based on
/// PieceToHistory instead of ButterflyBoards.
/// (~63 elo)
using ContinuationHistory = Stats<PieceToHistory, NOT_USED, PIECE_NB, SQUARE_NB>;
/// MovePicker class is used to pick one pseudo-legal move at a time from the
/// current position. The most important method is next_move(), which returns a
/// new pseudo-legal move each time it is called, until there are no moves left,
/// when MOVE_NONE is returned. In order to improve the efficiency of the
/// alpha-beta algorithm, MovePicker attempts to return the moves which are most
/// likely to get a cut-off first.
class MovePicker {
enum PickType { Next, Best };
public:
MovePicker(const MovePicker&) = delete;
MovePicker& operator=(const MovePicker&) = delete;
MovePicker(const Position&, Move, Depth, const ButterflyHistory*,
const CapturePieceToHistory*,
const PieceToHistory**,
Move,
const Move*);
MovePicker(const Position&, Move, Depth, const ButterflyHistory*,
const CapturePieceToHistory*,
const PieceToHistory**,
Square);
MovePicker(const Position&, Move, Value, const CapturePieceToHistory*);
Move next_move(bool skipQuiets = false);
private:
template<PickType T, typename Pred> Move select(Pred);
template<GenType> void score();
ExtMove* begin() { return cur; }
ExtMove* end() { return endMoves; }
const Position& pos;
const ButterflyHistory* mainHistory;
const CapturePieceToHistory* captureHistory;
const PieceToHistory** continuationHistory;
Move ttMove;
ExtMove refutations[3], *cur, *endMoves, *endBadCaptures;
int stage;
Square recaptureSquare;
Value threshold;
Depth depth;
ExtMove moves[MAX_MOVES];
};
} // namespace Stockfish
#endif // #ifndef MOVEPICK_H_INCLUDED

405
src/Stockfish/src/nnue/evaluate_nnue.cpp Normal file
View file

@ -0,0 +1,405 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// Code for calculating NNUE evaluation function
#include <fstream>
#include <iomanip>
#include <iostream>
#include <set>
#include <sstream>
#include <string_view>
#include "../evaluate.h"
#include "../position.h"
#include "../uci.h"
#include "../types.h"
#include "evaluate_nnue.h"
namespace Stockfish::Eval::NNUE {
// Input feature converter
LargePagePtr<FeatureTransformer> featureTransformer;
// Evaluation function
AlignedPtr<Network> network[LayerStacks];
// Evaluation function file name
std::string fileName;
std::string netDescription;
namespace Detail {
// Initialize the evaluation function parameters
template <typename T>
void initialize(AlignedPtr<T>& pointer) {
pointer.reset(reinterpret_cast<T*>(std_aligned_alloc(alignof(T), sizeof(T))));
std::memset(pointer.get(), 0, sizeof(T));
}
template <typename T>
void initialize(LargePagePtr<T>& pointer) {
static_assert(alignof(T) <= 4096, "aligned_large_pages_alloc() may fail for such a big alignment requirement of T");
pointer.reset(reinterpret_cast<T*>(aligned_large_pages_alloc(sizeof(T))));
std::memset(pointer.get(), 0, sizeof(T));
}
// Read evaluation function parameters
template <typename T>
bool read_parameters(std::istream& stream, T& reference) {
std::uint32_t header;
header = read_little_endian<std::uint32_t>(stream);
if (!stream || header != T::get_hash_value()) return false;
return reference.read_parameters(stream);
}
// Write evaluation function parameters
template <typename T>
bool write_parameters(std::ostream& stream, const T& reference) {
write_little_endian<std::uint32_t>(stream, T::get_hash_value());
return reference.write_parameters(stream);
}
} // namespace Detail
// Initialize the evaluation function parameters
static void initialize() {
Detail::initialize(featureTransformer);
for (std::size_t i = 0; i < LayerStacks; ++i)
Detail::initialize(network[i]);
}
// Read network header
static bool read_header(std::istream& stream, std::uint32_t* hashValue, std::string* desc)
{
std::uint32_t version, size;
version = read_little_endian<std::uint32_t>(stream);
*hashValue = read_little_endian<std::uint32_t>(stream);
size = read_little_endian<std::uint32_t>(stream);
if (!stream || version != Version) return false;
desc->resize(size);
stream.read(&(*desc)[0], size);
return !stream.fail();
}
// Write network header
static bool write_header(std::ostream& stream, std::uint32_t hashValue, const std::string& desc)
{
write_little_endian<std::uint32_t>(stream, Version);
write_little_endian<std::uint32_t>(stream, hashValue);
write_little_endian<std::uint32_t>(stream, (std::uint32_t)desc.size());
stream.write(&desc[0], desc.size());
return !stream.fail();
}
// Read network parameters
static bool read_parameters(std::istream& stream) {
std::uint32_t hashValue;
if (!read_header(stream, &hashValue, &netDescription)) return false;
if (hashValue != HashValue) return false;
if (!Detail::read_parameters(stream, *featureTransformer)) return false;
for (std::size_t i = 0; i < LayerStacks; ++i)
if (!Detail::read_parameters(stream, *(network[i]))) return false;
return stream && stream.peek() == std::ios::traits_type::eof();
}
// Write network parameters
static bool write_parameters(std::ostream& stream) {
if (!write_header(stream, HashValue, netDescription)) return false;
if (!Detail::write_parameters(stream, *featureTransformer)) return false;
for (std::size_t i = 0; i < LayerStacks; ++i)
if (!Detail::write_parameters(stream, *(network[i]))) return false;
return (bool)stream;
}
void hint_common_parent_position(const Position& pos) {
if (Eval::useNNUE)
featureTransformer->hint_common_access(pos);
}
// Evaluation function. Perform differential calculation.
Value evaluate(const Position& pos, bool adjusted, int* complexity) {
// We manually align the arrays on the stack because with gcc < 9.3
// overaligning stack variables with alignas() doesn't work correctly.
constexpr uint64_t alignment = CacheLineSize;
constexpr int delta = 24;
#if defined(ALIGNAS_ON_STACK_VARIABLES_BROKEN)
TransformedFeatureType transformedFeaturesUnaligned[
FeatureTransformer::BufferSize + alignment / sizeof(TransformedFeatureType)];
auto* transformedFeatures = align_ptr_up<alignment>(&transformedFeaturesUnaligned[0]);
#else
alignas(alignment)
TransformedFeatureType transformedFeatures[FeatureTransformer::BufferSize];
#endif
ASSERT_ALIGNED(transformedFeatures, alignment);
const int bucket = (pos.count<ALL_PIECES>() - 1) / 4;
const auto psqt = featureTransformer->transform(pos, transformedFeatures, bucket);
const auto positional = network[bucket]->propagate(transformedFeatures);
if (complexity)
*complexity = abs(psqt - positional) / OutputScale;
// Give more value to positional evaluation when adjusted flag is set
if (adjusted)
return static_cast<Value>(((1024 - delta) * psqt + (1024 + delta) * positional) / (1024 * OutputScale));
else
return static_cast<Value>((psqt + positional) / OutputScale);
}
struct NnueEvalTrace {
static_assert(LayerStacks == PSQTBuckets);
Value psqt[LayerStacks];
Value positional[LayerStacks];
std::size_t correctBucket;
};
static NnueEvalTrace trace_evaluate(const Position& pos) {
// We manually align the arrays on the stack because with gcc < 9.3
// overaligning stack variables with alignas() doesn't work correctly.
constexpr uint64_t alignment = CacheLineSize;
#if defined(ALIGNAS_ON_STACK_VARIABLES_BROKEN)
TransformedFeatureType transformedFeaturesUnaligned[
FeatureTransformer::BufferSize + alignment / sizeof(TransformedFeatureType)];
auto* transformedFeatures = align_ptr_up<alignment>(&transformedFeaturesUnaligned[0]);
#else
alignas(alignment)
TransformedFeatureType transformedFeatures[FeatureTransformer::BufferSize];
#endif
ASSERT_ALIGNED(transformedFeatures, alignment);
NnueEvalTrace t{};
t.correctBucket = (pos.count<ALL_PIECES>() - 1) / 4;
for (IndexType bucket = 0; bucket < LayerStacks; ++bucket) {
const auto materialist = featureTransformer->transform(pos, transformedFeatures, bucket);
const auto positional = network[bucket]->propagate(transformedFeatures);
t.psqt[bucket] = static_cast<Value>( materialist / OutputScale );
t.positional[bucket] = static_cast<Value>( positional / OutputScale );
}
return t;
}
constexpr std::string_view PieceToChar(" PNBRQK pnbrqk");
// format_cp_compact() converts a Value into (centi)pawns and writes it in a buffer.
// The buffer must have capacity for at least 5 chars.
static void format_cp_compact(Value v, char* buffer) {
buffer[0] = (v < 0 ? '-' : v > 0 ? '+' : ' ');
int cp = std::abs(100 * v / UCI::NormalizeToPawnValue);
if (cp >= 10000)
{
buffer[1] = '0' + cp / 10000; cp %= 10000;
buffer[2] = '0' + cp / 1000; cp %= 1000;
buffer[3] = '0' + cp / 100;
buffer[4] = ' ';
}
else if (cp >= 1000)
{
buffer[1] = '0' + cp / 1000; cp %= 1000;
buffer[2] = '0' + cp / 100; cp %= 100;
buffer[3] = '.';
buffer[4] = '0' + cp / 10;
}
else
{
buffer[1] = '0' + cp / 100; cp %= 100;
buffer[2] = '.';
buffer[3] = '0' + cp / 10; cp %= 10;
buffer[4] = '0' + cp / 1;
}
}
// format_cp_aligned_dot() converts a Value into (centi)pawns, always keeping two decimals.
static void format_cp_aligned_dot(Value v, std::stringstream &stream) {
const double cp = 1.0 * std::abs(int(v)) / UCI::NormalizeToPawnValue;
stream << (v < 0 ? '-' : v > 0 ? '+' : ' ')
<< std::setiosflags(std::ios::fixed)
<< std::setw(6)
<< std::setprecision(2)
<< cp;
}
// trace() returns a string with the value of each piece on a board,
// and a table for (PSQT, Layers) values bucket by bucket.
std::string trace(Position& pos) {
std::stringstream ss;
char board[3*8+1][8*8+2];
std::memset(board, ' ', sizeof(board));
for (int row = 0; row < 3*8+1; ++row)
board[row][8*8+1] = '\0';
// A lambda to output one box of the board
auto writeSquare = [&board](File file, Rank rank, Piece pc, Value value) {
const int x = ((int)file) * 8;
const int y = (7 - (int)rank) * 3;
for (int i = 1; i < 8; ++i)
board[y][x+i] = board[y+3][x+i] = '-';
for (int i = 1; i < 3; ++i)
board[y+i][x] = board[y+i][x+8] = '|';
board[y][x] = board[y][x+8] = board[y+3][x+8] = board[y+3][x] = '+';
if (pc != NO_PIECE)
board[y+1][x+4] = PieceToChar[pc];
if (value != VALUE_NONE)
format_cp_compact(value, &board[y+2][x+2]);
};
// We estimate the value of each piece by doing a differential evaluation from
// the current base eval, simulating the removal of the piece from its square.
Value base = evaluate(pos);
base = pos.side_to_move() == WHITE ? base : -base;
for (File f = FILE_A; f <= FILE_H; ++f)
for (Rank r = RANK_1; r <= RANK_8; ++r)
{
Square sq = make_square(f, r);
Piece pc = pos.piece_on(sq);
Value v = VALUE_NONE;
if (pc != NO_PIECE && type_of(pc) != KING)
{
auto st = pos.state();
pos.remove_piece(sq);
st->accumulator.computed[WHITE] = false;
st->accumulator.computed[BLACK] = false;
Value eval = evaluate(pos);
eval = pos.side_to_move() == WHITE ? eval : -eval;
v = base - eval;
pos.put_piece(pc, sq);
st->accumulator.computed[WHITE] = false;
st->accumulator.computed[BLACK] = false;
}
writeSquare(f, r, pc, v);
}
ss << " NNUE derived piece values:\n";
for (int row = 0; row < 3*8+1; ++row)
ss << board[row] << '\n';
ss << '\n';
auto t = trace_evaluate(pos);
ss << " NNUE network contributions "
<< (pos.side_to_move() == WHITE ? "(White to move)" : "(Black to move)") << fakeendl
<< "+------------+------------+------------+------------+\n"
<< "| Bucket | Material | Positional | Total |\n"
<< "| | (PSQT) | (Layers) | |\n"
<< "+------------+------------+------------+------------+\n";
for (std::size_t bucket = 0; bucket < LayerStacks; ++bucket)
{
ss << "| " << bucket << " ";
ss << " | "; format_cp_aligned_dot(t.psqt[bucket], ss); ss << " "
<< " | "; format_cp_aligned_dot(t.positional[bucket], ss); ss << " "
<< " | "; format_cp_aligned_dot(t.psqt[bucket] + t.positional[bucket], ss); ss << " "
<< " |";
if (bucket == t.correctBucket)
ss << " <-- this bucket is used";
ss << '\n';
}
ss << "+------------+------------+------------+------------+\n";
return ss.str();
}
// Load eval, from a file stream or a memory stream
bool load_eval(std::string name, std::istream& stream) {
initialize();
fileName = name;
return read_parameters(stream);
}
// Save eval, to a file stream or a memory stream
bool save_eval(std::ostream& stream) {
if (fileName.empty())
return false;
return write_parameters(stream);
}
/// Save eval, to a file given by its name
bool save_eval(const std::optional<std::string>& filename) {
std::string actualFilename;
std::string msg;
if (filename.has_value())
actualFilename = filename.value();
else
{
if (currentEvalFileName != EvalFileDefaultName)
{
msg = "Failed to export a net. A non-embedded net can only be saved if the filename is specified";
sync_cout << msg << sync_endl;
return false;
}
actualFilename = EvalFileDefaultName;
}
std::ofstream stream(actualFilename, std::ios_base::binary);
bool saved = save_eval(stream);
msg = saved ? "Network saved successfully to " + actualFilename
: "Failed to export a net";
sync_cout << msg << sync_endl;
return saved;
}
} // namespace Stockfish::Eval::NNUE

68
src/Stockfish/src/nnue/evaluate_nnue.h Normal file
View file

@ -0,0 +1,68 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// header used in NNUE evaluation function
#ifndef NNUE_EVALUATE_NNUE_H_INCLUDED
#define NNUE_EVALUATE_NNUE_H_INCLUDED
#include "nnue_feature_transformer.h"
#include <memory>
namespace Stockfish::Eval::NNUE {
// Hash value of evaluation function structure
constexpr std::uint32_t HashValue =
FeatureTransformer::get_hash_value() ^ Network::get_hash_value();
// Deleter for automating release of memory area
template <typename T>
struct AlignedDeleter {
void operator()(T* ptr) const {
ptr->~T();
std_aligned_free(ptr);
}
};
template <typename T>
struct LargePageDeleter {
void operator()(T* ptr) const {
ptr->~T();
aligned_large_pages_free(ptr);
}
};
template <typename T>
using AlignedPtr = std::unique_ptr<T, AlignedDeleter<T>>;
template <typename T>
using LargePagePtr = std::unique_ptr<T, LargePageDeleter<T>>;
std::string trace(Position& pos);
Value evaluate(const Position& pos, bool adjusted = false, int* complexity = nullptr);
void hint_common_parent_position(const Position& pos);
bool load_eval(std::string name, std::istream& stream);
bool save_eval(std::ostream& stream);
bool save_eval(const std::optional<std::string>& filename);
} // namespace Stockfish::Eval::NNUE
#endif // #ifndef NNUE_EVALUATE_NNUE_H_INCLUDED

84
src/Stockfish/src/nnue/features/half_ka_v2_hm.cpp Normal file
View file

@ -0,0 +1,84 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
//Definition of input features HalfKAv2_hm of NNUE evaluation function
#include "half_ka_v2_hm.h"
#include "../../position.h"
namespace Stockfish::Eval::NNUE::Features {
// Index of a feature for a given king position and another piece on some square
template<Color Perspective>
inline IndexType HalfKAv2_hm::make_index(Square s, Piece pc, Square ksq) {
return IndexType((int(s) ^ OrientTBL[Perspective][ksq]) + PieceSquareIndex[Perspective][pc] + KingBuckets[Perspective][ksq]);
}
// Get a list of indices for active features
template<Color Perspective>
void HalfKAv2_hm::append_active_indices(
const Position& pos,
IndexList& active
) {
Square ksq = pos.square<KING>(Perspective);
Bitboard bb = pos.pieces();
while (bb)
{
Square s = pop_lsb(bb);
active.push_back(make_index<Perspective>(s, pos.piece_on(s), ksq));
}
}
// Explicit template instantiations
template void HalfKAv2_hm::append_active_indices<WHITE>(const Position& pos, IndexList& active);
template void HalfKAv2_hm::append_active_indices<BLACK>(const Position& pos, IndexList& active);
// append_changed_indices() : get a list of indices for recently changed features
template<Color Perspective>
void HalfKAv2_hm::append_changed_indices(
Square ksq,
const DirtyPiece& dp,
IndexList& removed,
IndexList& added
) {
for (int i = 0; i < dp.dirty_num; ++i) {
if (dp.from[i] != SQ_NONE)
removed.push_back(make_index<Perspective>(dp.from[i], dp.piece[i], ksq));
if (dp.to[i] != SQ_NONE)
added.push_back(make_index<Perspective>(dp.to[i], dp.piece[i], ksq));
}
}
// Explicit template instantiations
template void HalfKAv2_hm::append_changed_indices<WHITE>(Square ksq, const DirtyPiece& dp, IndexList& removed, IndexList& added);
template void HalfKAv2_hm::append_changed_indices<BLACK>(Square ksq, const DirtyPiece& dp, IndexList& removed, IndexList& added);
int HalfKAv2_hm::update_cost(const StateInfo* st) {
return st->dirtyPiece.dirty_num;
}
int HalfKAv2_hm::refresh_cost(const Position& pos) {
return pos.count<ALL_PIECES>();
}
bool HalfKAv2_hm::requires_refresh(const StateInfo* st, Color perspective) {
return st->dirtyPiece.piece[0] == make_piece(perspective, KING);
}
} // namespace Stockfish::Eval::NNUE::Features

152
src/Stockfish/src/nnue/features/half_ka_v2_hm.h Normal file
View file

@ -0,0 +1,152 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
//Definition of input features HalfKP of NNUE evaluation function
#ifndef NNUE_FEATURES_HALF_KA_V2_HM_H_INCLUDED
#define NNUE_FEATURES_HALF_KA_V2_HM_H_INCLUDED
#include "../nnue_common.h"
#include "../../evaluate.h"
#include "../../misc.h"
namespace Stockfish {
struct StateInfo;
}
namespace Stockfish::Eval::NNUE::Features {
// Feature HalfKAv2_hm: Combination of the position of own king
// and the position of pieces. Position mirrored such that king always on e..h files.
class HalfKAv2_hm {
// unique number for each piece type on each square
enum {
PS_NONE = 0,
PS_W_PAWN = 0,
PS_B_PAWN = 1 * SQUARE_NB,
PS_W_KNIGHT = 2 * SQUARE_NB,
PS_B_KNIGHT = 3 * SQUARE_NB,
PS_W_BISHOP = 4 * SQUARE_NB,
PS_B_BISHOP = 5 * SQUARE_NB,
PS_W_ROOK = 6 * SQUARE_NB,
PS_B_ROOK = 7 * SQUARE_NB,
PS_W_QUEEN = 8 * SQUARE_NB,
PS_B_QUEEN = 9 * SQUARE_NB,
PS_KING = 10 * SQUARE_NB,
PS_NB = 11 * SQUARE_NB
};
static constexpr IndexType PieceSquareIndex[COLOR_NB][PIECE_NB] = {
// convention: W - us, B - them
// viewed from other side, W and B are reversed
{ PS_NONE, PS_W_PAWN, PS_W_KNIGHT, PS_W_BISHOP, PS_W_ROOK, PS_W_QUEEN, PS_KING, PS_NONE,
PS_NONE, PS_B_PAWN, PS_B_KNIGHT, PS_B_BISHOP, PS_B_ROOK, PS_B_QUEEN, PS_KING, PS_NONE },
{ PS_NONE, PS_B_PAWN, PS_B_KNIGHT, PS_B_BISHOP, PS_B_ROOK, PS_B_QUEEN, PS_KING, PS_NONE,
PS_NONE, PS_W_PAWN, PS_W_KNIGHT, PS_W_BISHOP, PS_W_ROOK, PS_W_QUEEN, PS_KING, PS_NONE }
};
// Index of a feature for a given king position and another piece on some square
template<Color Perspective>
static IndexType make_index(Square s, Piece pc, Square ksq);
public:
// Feature name
static constexpr const char* Name = "HalfKAv2_hm(Friend)";
// Hash value embedded in the evaluation file
static constexpr std::uint32_t HashValue = 0x7f234cb8u;
// Number of feature dimensions
static constexpr IndexType Dimensions =
static_cast<IndexType>(SQUARE_NB) * static_cast<IndexType>(PS_NB) / 2;
#define B(v) (v * PS_NB)
static constexpr int KingBuckets[COLOR_NB][SQUARE_NB] = {
{ B(28), B(29), B(30), B(31), B(31), B(30), B(29), B(28),
B(24), B(25), B(26), B(27), B(27), B(26), B(25), B(24),
B(20), B(21), B(22), B(23), B(23), B(22), B(21), B(20),
B(16), B(17), B(18), B(19), B(19), B(18), B(17), B(16),
B(12), B(13), B(14), B(15), B(15), B(14), B(13), B(12),
B( 8), B( 9), B(10), B(11), B(11), B(10), B( 9), B( 8),
B( 4), B( 5), B( 6), B( 7), B( 7), B( 6), B( 5), B( 4),
B( 0), B( 1), B( 2), B( 3), B( 3), B( 2), B( 1), B( 0) },
{ B( 0), B( 1), B( 2), B( 3), B( 3), B( 2), B( 1), B( 0),
B( 4), B( 5), B( 6), B( 7), B( 7), B( 6), B( 5), B( 4),
B( 8), B( 9), B(10), B(11), B(11), B(10), B( 9), B( 8),
B(12), B(13), B(14), B(15), B(15), B(14), B(13), B(12),
B(16), B(17), B(18), B(19), B(19), B(18), B(17), B(16),
B(20), B(21), B(22), B(23), B(23), B(22), B(21), B(20),
B(24), B(25), B(26), B(27), B(27), B(26), B(25), B(24),
B(28), B(29), B(30), B(31), B(31), B(30), B(29), B(28) }
};
#undef B
// Orient a square according to perspective (rotates by 180 for black)
static constexpr int OrientTBL[COLOR_NB][SQUARE_NB] = {
{ SQ_H1, SQ_H1, SQ_H1, SQ_H1, SQ_A1, SQ_A1, SQ_A1, SQ_A1,
SQ_H1, SQ_H1, SQ_H1, SQ_H1, SQ_A1, SQ_A1, SQ_A1, SQ_A1,
SQ_H1, SQ_H1, SQ_H1, SQ_H1, SQ_A1, SQ_A1, SQ_A1, SQ_A1,
SQ_H1, SQ_H1, SQ_H1, SQ_H1, SQ_A1, SQ_A1, SQ_A1, SQ_A1,
SQ_H1, SQ_H1, SQ_H1, SQ_H1, SQ_A1, SQ_A1, SQ_A1, SQ_A1,
SQ_H1, SQ_H1, SQ_H1, SQ_H1, SQ_A1, SQ_A1, SQ_A1, SQ_A1,
SQ_H1, SQ_H1, SQ_H1, SQ_H1, SQ_A1, SQ_A1, SQ_A1, SQ_A1,
SQ_H1, SQ_H1, SQ_H1, SQ_H1, SQ_A1, SQ_A1, SQ_A1, SQ_A1 },
{ SQ_H8, SQ_H8, SQ_H8, SQ_H8, SQ_A8, SQ_A8, SQ_A8, SQ_A8,
SQ_H8, SQ_H8, SQ_H8, SQ_H8, SQ_A8, SQ_A8, SQ_A8, SQ_A8,
SQ_H8, SQ_H8, SQ_H8, SQ_H8, SQ_A8, SQ_A8, SQ_A8, SQ_A8,
SQ_H8, SQ_H8, SQ_H8, SQ_H8, SQ_A8, SQ_A8, SQ_A8, SQ_A8,
SQ_H8, SQ_H8, SQ_H8, SQ_H8, SQ_A8, SQ_A8, SQ_A8, SQ_A8,
SQ_H8, SQ_H8, SQ_H8, SQ_H8, SQ_A8, SQ_A8, SQ_A8, SQ_A8,
SQ_H8, SQ_H8, SQ_H8, SQ_H8, SQ_A8, SQ_A8, SQ_A8, SQ_A8,
SQ_H8, SQ_H8, SQ_H8, SQ_H8, SQ_A8, SQ_A8, SQ_A8, SQ_A8 }
};
// Maximum number of simultaneously active features.
static constexpr IndexType MaxActiveDimensions = 32;
using IndexList = ValueList<IndexType, MaxActiveDimensions>;
// Get a list of indices for active features
template<Color Perspective>
static void append_active_indices(
const Position& pos,
IndexList& active);
// Get a list of indices for recently changed features
template<Color Perspective>
static void append_changed_indices(
Square ksq,
const DirtyPiece& dp,
IndexList& removed,
IndexList& added
);
// Returns the cost of updating one perspective, the most costly one.
// Assumes no refresh needed.
static int update_cost(const StateInfo* st);
static int refresh_cost(const Position& pos);
// Returns whether the change stored in this StateInfo means that
// a full accumulator refresh is required.
static bool requires_refresh(const StateInfo* st, Color perspective);
};
} // namespace Stockfish::Eval::NNUE::Features
#endif // #ifndef NNUE_FEATURES_HALF_KA_V2_HM_H_INCLUDED

563
src/Stockfish/src/nnue/layers/affine_transform.h Normal file
View file

@ -0,0 +1,563 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// Definition of layer AffineTransform of NNUE evaluation function
#ifndef NNUE_LAYERS_AFFINE_TRANSFORM_H_INCLUDED
#define NNUE_LAYERS_AFFINE_TRANSFORM_H_INCLUDED
#include <iostream>
#include <algorithm>
#include <type_traits>
#include "../nnue_common.h"
#include "simd.h"
/*
This file contains the definition for a fully connected layer (aka affine transform).
Two approaches are employed, depending on the sizes of the transform.
Approach 1 (a specialization for large inputs):
- used when the PaddedInputDimensions >= 128
- uses AVX512 if possible
- processes inputs in batches of 2*InputSimdWidth
- so in batches of 128 for AVX512
- the weight blocks of size InputSimdWidth are transposed such that
access is sequential
- N columns of the weight matrix are processed a time, where N
depends on the architecture (the amount of registers)
- accumulate + hadd is used
Approach 2 (a specialization for small inputs):
- used when the PaddedInputDimensions < 128
- expected use-case is for when PaddedInputDimensions == 32 and InputDimensions <= 32.
- that's why AVX512 is hard to implement
- expected use-case is small layers
- not optimized as well as the approach 1
- inputs are processed in chunks of 4, weights are respectively transposed
- accumulation happens directly to int32s
*/
namespace Stockfish::Eval::NNUE::Layers {
// Fallback implementation for older/other architectures.
// Identical for both approaches. Requires the input to be padded to at least 16 values.
#if !defined(USE_SSSE3)
template <IndexType InputDimensions, IndexType PaddedInputDimensions, IndexType OutputDimensions>
static void affine_transform_non_ssse3(std::int32_t* output, const std::int8_t* weights, const std::int32_t* biases, const std::uint8_t* input)
{
# if defined(USE_SSE2)
// At least a multiple of 16, with SSE2.
constexpr IndexType NumChunks = ceil_to_multiple<IndexType>(InputDimensions, 16) / 16;
const __m128i Zeros = _mm_setzero_si128();
const auto inputVector = reinterpret_cast<const __m128i*>(input);
# elif defined(USE_MMX)
constexpr IndexType NumChunks = ceil_to_multiple<IndexType>(InputDimensions, 8) / 8;
const __m64 Zeros = _mm_setzero_si64();
const auto inputVector = reinterpret_cast<const __m64*>(input);
# elif defined(USE_NEON_DOTPROD)
constexpr IndexType NumChunks = ceil_to_multiple<IndexType>(InputDimensions, 16) / 16;
const auto inputVector = reinterpret_cast<const int8x16_t*>(input);
# elif defined(USE_NEON)
constexpr IndexType NumChunks = ceil_to_multiple<IndexType>(InputDimensions, 16) / 16;
const auto inputVector = reinterpret_cast<const int8x8_t*>(input);
# endif
for (IndexType i = 0; i < OutputDimensions; ++i) {
const IndexType offset = i * PaddedInputDimensions;
# if defined(USE_SSE2)
__m128i sumLo = _mm_cvtsi32_si128(biases[i]);
__m128i sumHi = Zeros;
const auto row = reinterpret_cast<const __m128i*>(&weights[offset]);
for (IndexType j = 0; j < NumChunks; ++j) {
__m128i row_j = _mm_load_si128(&row[j]);
__m128i input_j = _mm_load_si128(&inputVector[j]);
__m128i extendedRowLo = _mm_srai_epi16(_mm_unpacklo_epi8(row_j, row_j), 8);
__m128i extendedRowHi = _mm_srai_epi16(_mm_unpackhi_epi8(row_j, row_j), 8);
__m128i extendedInputLo = _mm_unpacklo_epi8(input_j, Zeros);
__m128i extendedInputHi = _mm_unpackhi_epi8(input_j, Zeros);
__m128i productLo = _mm_madd_epi16(extendedRowLo, extendedInputLo);
__m128i productHi = _mm_madd_epi16(extendedRowHi, extendedInputHi);
sumLo = _mm_add_epi32(sumLo, productLo);
sumHi = _mm_add_epi32(sumHi, productHi);
}
__m128i sum = _mm_add_epi32(sumLo, sumHi);
__m128i sumHigh_64 = _mm_shuffle_epi32(sum, _MM_SHUFFLE(1, 0, 3, 2));
sum = _mm_add_epi32(sum, sumHigh_64);
__m128i sum_second_32 = _mm_shufflelo_epi16(sum, _MM_SHUFFLE(1, 0, 3, 2));
sum = _mm_add_epi32(sum, sum_second_32);
output[i] = _mm_cvtsi128_si32(sum);
# elif defined(USE_MMX)
__m64 sumLo = _mm_cvtsi32_si64(biases[i]);
__m64 sumHi = Zeros;
const auto row = reinterpret_cast<const __m64*>(&weights[offset]);
for (IndexType j = 0; j < NumChunks; ++j) {
__m64 row_j = row[j];
__m64 input_j = inputVector[j];
__m64 extendedRowLo = _mm_srai_pi16(_mm_unpacklo_pi8(row_j, row_j), 8);
__m64 extendedRowHi = _mm_srai_pi16(_mm_unpackhi_pi8(row_j, row_j), 8);
__m64 extendedInputLo = _mm_unpacklo_pi8(input_j, Zeros);
__m64 extendedInputHi = _mm_unpackhi_pi8(input_j, Zeros);
__m64 productLo = _mm_madd_pi16(extendedRowLo, extendedInputLo);
__m64 productHi = _mm_madd_pi16(extendedRowHi, extendedInputHi);
sumLo = _mm_add_pi32(sumLo, productLo);
sumHi = _mm_add_pi32(sumHi, productHi);
}
__m64 sum = _mm_add_pi32(sumLo, sumHi);
sum = _mm_add_pi32(sum, _mm_unpackhi_pi32(sum, sum));
output[i] = _mm_cvtsi64_si32(sum);
# elif defined(USE_NEON_DOTPROD)
int32x4_t sum = {biases[i]};
const auto row = reinterpret_cast<const int8x16_t*>(&weights[offset]);
for (IndexType j = 0; j < NumChunks; ++j) {
sum = vdotq_s32(sum, inputVector[j], row[j]);
}
output[i] = vaddvq_s32(sum);
# elif defined(USE_NEON)
int32x4_t sum = {biases[i]};
const auto row = reinterpret_cast<const int8x8_t*>(&weights[offset]);
for (IndexType j = 0; j < NumChunks; ++j) {
int16x8_t product = vmull_s8(inputVector[j * 2], row[j * 2]);
product = vmlal_s8(product, inputVector[j * 2 + 1], row[j * 2 + 1]);
sum = vpadalq_s16(sum, product);
}
output[i] = sum[0] + sum[1] + sum[2] + sum[3];
# else
std::int32_t sum = biases[i];
for (IndexType j = 0; j < InputDimensions; ++j) {
sum += weights[offset + j] * input[j];
}
output[i] = sum;
# endif
}
# if defined(USE_MMX)
_mm_empty();
# endif
}
#endif
template <IndexType InDims, IndexType OutDims, typename Enabled = void>
class AffineTransform;
#if defined (USE_AVX512)
constexpr IndexType LargeInputSize = 2 * 64;
#else
constexpr IndexType LargeInputSize = std::numeric_limits<IndexType>::max();
#endif
// A specialization for large inputs
template <IndexType InDims, IndexType OutDims>
class AffineTransform<InDims, OutDims, std::enable_if_t<(ceil_to_multiple<IndexType>(InDims, MaxSimdWidth) >= LargeInputSize)>> {
public:
// Input/output type
using InputType = std::uint8_t;
using OutputType = std::int32_t;
// Number of input/output dimensions
static constexpr IndexType InputDimensions = InDims;
static constexpr IndexType OutputDimensions = OutDims;
static constexpr IndexType PaddedInputDimensions =
ceil_to_multiple<IndexType>(InputDimensions, MaxSimdWidth);
static constexpr IndexType PaddedOutputDimensions =
ceil_to_multiple<IndexType>(OutputDimensions, MaxSimdWidth);
using OutputBuffer = OutputType[PaddedOutputDimensions];
static_assert(PaddedInputDimensions >= LargeInputSize, "Something went wrong. This specialization (for large inputs) should not have been chosen.");
#if defined (USE_AVX512)
static constexpr IndexType InputSimdWidth = 64;
static constexpr IndexType MaxNumOutputRegs = 16;
#elif defined (USE_AVX2)
static constexpr IndexType InputSimdWidth = 32;
static constexpr IndexType MaxNumOutputRegs = 8;
#elif defined (USE_SSSE3)
static constexpr IndexType InputSimdWidth = 16;
static constexpr IndexType MaxNumOutputRegs = 8;
#elif defined (USE_NEON_DOTPROD)
static constexpr IndexType InputSimdWidth = 16;
static constexpr IndexType MaxNumOutputRegs = 8;
#elif defined (USE_NEON)
static constexpr IndexType InputSimdWidth = 8;
static constexpr IndexType MaxNumOutputRegs = 8;
#else
// The fallback implementation will not have permuted weights.
// We define these to avoid a lot of ifdefs later.
static constexpr IndexType InputSimdWidth = 1;
static constexpr IndexType MaxNumOutputRegs = 1;
#endif
// A big block is a region in the weight matrix of the size [PaddedInputDimensions, NumOutputRegs].
// A small block is a region of size [InputSimdWidth, 1]
static constexpr IndexType NumOutputRegs = std::min(MaxNumOutputRegs, OutputDimensions);
static constexpr IndexType SmallBlockSize = InputSimdWidth;
static constexpr IndexType BigBlockSize = NumOutputRegs * PaddedInputDimensions;
static constexpr IndexType NumSmallBlocksInBigBlock = BigBlockSize / SmallBlockSize;
static constexpr IndexType NumSmallBlocksPerOutput = PaddedInputDimensions / SmallBlockSize;
static constexpr IndexType NumBigBlocks = OutputDimensions / NumOutputRegs;
static_assert(OutputDimensions % NumOutputRegs == 0);
// Hash value embedded in the evaluation file
static constexpr std::uint32_t get_hash_value(std::uint32_t prevHash) {
std::uint32_t hashValue = 0xCC03DAE4u;
hashValue += OutputDimensions;
hashValue ^= prevHash >> 1;
hashValue ^= prevHash << 31;
return hashValue;
}
/*
Transposes the small blocks within a block.
Effectively means that weights can be traversed sequentially during inference.
*/
static IndexType get_weight_index(IndexType i)
{
const IndexType smallBlock = (i / SmallBlockSize) % NumSmallBlocksInBigBlock;
const IndexType smallBlockCol = smallBlock / NumSmallBlocksPerOutput;
const IndexType smallBlockRow = smallBlock % NumSmallBlocksPerOutput;
const IndexType bigBlock = i / BigBlockSize;
const IndexType rest = i % SmallBlockSize;
const IndexType idx =
bigBlock * BigBlockSize
+ smallBlockRow * SmallBlockSize * NumOutputRegs
+ smallBlockCol * SmallBlockSize
+ rest;
return idx;
}
// Read network parameters
bool read_parameters(std::istream& stream) {
read_little_endian<BiasType>(stream, biases, OutputDimensions);
for (IndexType i = 0; i < OutputDimensions * PaddedInputDimensions; ++i)
weights[get_weight_index(i)] = read_little_endian<WeightType>(stream);
return !stream.fail();
}
// Write network parameters
bool write_parameters(std::ostream& stream) const {
write_little_endian<BiasType>(stream, biases, OutputDimensions);
for (IndexType i = 0; i < OutputDimensions * PaddedInputDimensions; ++i)
write_little_endian<WeightType>(stream, weights[get_weight_index(i)]);
return !stream.fail();
}
// Forward propagation
const OutputType* propagate(
const InputType* input, OutputType* output) const {
#if defined (USE_AVX512)
using acc_vec_t = __m512i;
using bias_vec_t = __m128i;
using weight_vec_t = __m512i;
using in_vec_t = __m512i;
#define vec_zero _mm512_setzero_si512()
#define vec_add_dpbusd_32x2 Simd::m512_add_dpbusd_epi32x2
#define vec_hadd Simd::m512_hadd
#define vec_haddx4 Simd::m512_haddx4
#elif defined (USE_AVX2)
using acc_vec_t = __m256i;
using bias_vec_t = __m128i;
using weight_vec_t = __m256i;
using in_vec_t = __m256i;
#define vec_zero _mm256_setzero_si256()
#define vec_add_dpbusd_32x2 Simd::m256_add_dpbusd_epi32x2
#define vec_hadd Simd::m256_hadd
#define vec_haddx4 Simd::m256_haddx4
#elif defined (USE_SSSE3)
using acc_vec_t = __m128i;
using bias_vec_t = __m128i;
using weight_vec_t = __m128i;
using in_vec_t = __m128i;
#define vec_zero _mm_setzero_si128()
#define vec_add_dpbusd_32x2 Simd::m128_add_dpbusd_epi32x2
#define vec_hadd Simd::m128_hadd
#define vec_haddx4 Simd::m128_haddx4
#elif defined (USE_NEON_DOTPROD)
using acc_vec_t = int32x4_t;
using bias_vec_t = int32x4_t;
using weight_vec_t = int8x16_t;
using in_vec_t = int8x16_t;
#define vec_zero {0}
#define vec_add_dpbusd_32x2 Simd::dotprod_m128_add_dpbusd_epi32x2
#define vec_hadd Simd::neon_m128_hadd
#define vec_haddx4 Simd::neon_m128_haddx4
#elif defined (USE_NEON)
using acc_vec_t = int32x4_t;
using bias_vec_t = int32x4_t;
using weight_vec_t = int8x8_t;
using in_vec_t = int8x8_t;
#define vec_zero {0}
#define vec_add_dpbusd_32x2 Simd::neon_m128_add_dpbusd_epi32x2
#define vec_hadd Simd::neon_m128_hadd
#define vec_haddx4 Simd::neon_m128_haddx4
#endif
#if defined (USE_SSSE3) || defined (USE_NEON)
const in_vec_t* invec = reinterpret_cast<const in_vec_t*>(input);
// Perform accumulation to registers for each big block
for (IndexType bigBlock = 0; bigBlock < NumBigBlocks; ++bigBlock)
{
acc_vec_t acc[NumOutputRegs] = { vec_zero };
// Each big block has NumOutputRegs small blocks in each "row", one per register.
// We process two small blocks at a time to save on one addition without VNNI.
for (IndexType smallBlock = 0; smallBlock < NumSmallBlocksPerOutput; smallBlock += 2)
{
const weight_vec_t* weightvec =
reinterpret_cast<const weight_vec_t*>(
weights
+ bigBlock * BigBlockSize
+ smallBlock * SmallBlockSize * NumOutputRegs);
const in_vec_t in0 = invec[smallBlock + 0];
const in_vec_t in1 = invec[smallBlock + 1];
for (IndexType k = 0; k < NumOutputRegs; ++k)
vec_add_dpbusd_32x2(acc[k], in0, weightvec[k], in1, weightvec[k + NumOutputRegs]);
}
// Horizontally add all accumulators.
if constexpr (NumOutputRegs % 4 == 0)
{
bias_vec_t* outputvec = reinterpret_cast<bias_vec_t*>(output);
const bias_vec_t* biasvec = reinterpret_cast<const bias_vec_t*>(biases);
for (IndexType k = 0; k < NumOutputRegs; k += 4)
{
const IndexType idx = (bigBlock * NumOutputRegs + k) / 4;
outputvec[idx] = vec_haddx4(acc[k+0], acc[k+1], acc[k+2], acc[k+3], biasvec[idx]);
}
}
else
{
for (IndexType k = 0; k < NumOutputRegs; ++k)
{
const IndexType idx = (bigBlock * NumOutputRegs + k);
output[idx] = vec_hadd(acc[k], biases[idx]);
}
}
}
# undef vec_zero
# undef vec_add_dpbusd_32x2
# undef vec_hadd
# undef vec_haddx4
#else
// Use old implementation for the other architectures.
affine_transform_non_ssse3<
InputDimensions,
PaddedInputDimensions,
OutputDimensions>(output, weights, biases, input);
#endif
return output;
}
private:
using BiasType = OutputType;
using WeightType = std::int8_t;
alignas(CacheLineSize) BiasType biases[OutputDimensions];
alignas(CacheLineSize) WeightType weights[OutputDimensions * PaddedInputDimensions];
};
// A specialization for small inputs
template <IndexType InDims, IndexType OutDims>
class AffineTransform<InDims, OutDims, std::enable_if_t<(ceil_to_multiple<IndexType>(InDims, MaxSimdWidth) < LargeInputSize)>> {
public:
// Input/output type
// Input/output type
using InputType = std::uint8_t;
using OutputType = std::int32_t;
// Number of input/output dimensions
static constexpr IndexType InputDimensions = InDims;
static constexpr IndexType OutputDimensions = OutDims;
static constexpr IndexType PaddedInputDimensions =
ceil_to_multiple<IndexType>(InputDimensions, MaxSimdWidth);
static constexpr IndexType PaddedOutputDimensions =
ceil_to_multiple<IndexType>(OutputDimensions, MaxSimdWidth);
using OutputBuffer = OutputType[PaddedOutputDimensions];
static_assert(PaddedInputDimensions < LargeInputSize, "Something went wrong. This specialization (for small inputs) should not have been chosen.");
// Hash value embedded in the evaluation file
static constexpr std::uint32_t get_hash_value(std::uint32_t prevHash) {
std::uint32_t hashValue = 0xCC03DAE4u;
hashValue += OutputDimensions;
hashValue ^= prevHash >> 1;
hashValue ^= prevHash << 31;
return hashValue;
}
static IndexType get_weight_index_scrambled(IndexType i)
{
return
(i / 4) % (PaddedInputDimensions / 4) * OutputDimensions * 4 +
i / PaddedInputDimensions * 4 +
i % 4;
}
static IndexType get_weight_index(IndexType i)
{
#if defined (USE_SSSE3)
return get_weight_index_scrambled(i);
#else
return i;
#endif
}
// Read network parameters
bool read_parameters(std::istream& stream) {
read_little_endian<BiasType>(stream, biases, OutputDimensions);
for (IndexType i = 0; i < OutputDimensions * PaddedInputDimensions; ++i)
weights[get_weight_index(i)] = read_little_endian<WeightType>(stream);
return !stream.fail();
}
// Write network parameters
bool write_parameters(std::ostream& stream) const {
write_little_endian<BiasType>(stream, biases, OutputDimensions);
for (IndexType i = 0; i < OutputDimensions * PaddedInputDimensions; ++i)
write_little_endian<WeightType>(stream, weights[get_weight_index(i)]);
return !stream.fail();
}
// Forward propagation
const OutputType* propagate(
const InputType* input, OutputType* output) const {
#if defined (USE_AVX512)
using vec_t = __m512i;
#define vec_setzero _mm512_setzero_si512
#define vec_set_32 _mm512_set1_epi32
#define vec_add_dpbusd_32 Simd::m512_add_dpbusd_epi32
#define vec_add_dpbusd_32x2 Simd::m512_add_dpbusd_epi32x2
#define vec_hadd Simd::m512_hadd
#elif defined (USE_AVX2)
using vec_t = __m256i;
#define vec_setzero _mm256_setzero_si256
#define vec_set_32 _mm256_set1_epi32
#define vec_add_dpbusd_32 Simd::m256_add_dpbusd_epi32
#define vec_add_dpbusd_32x2 Simd::m256_add_dpbusd_epi32x2
#define vec_hadd Simd::m256_hadd
#elif defined (USE_SSSE3)
using vec_t = __m128i;
#define vec_setzero _mm_setzero_si128
#define vec_set_32 _mm_set1_epi32
#define vec_add_dpbusd_32 Simd::m128_add_dpbusd_epi32
#define vec_add_dpbusd_32x2 Simd::m128_add_dpbusd_epi32x2
#define vec_hadd Simd::m128_hadd
#endif
#if defined (USE_SSSE3)
const auto inputVector = reinterpret_cast<const vec_t*>(input);
static constexpr IndexType OutputSimdWidth = sizeof(vec_t) / sizeof(OutputType);
static_assert(OutputDimensions % OutputSimdWidth == 0 || OutputDimensions == 1);
if constexpr (OutputDimensions % OutputSimdWidth == 0)
{
constexpr IndexType NumChunks = ceil_to_multiple<IndexType>(InputDimensions, 8) / 4;
constexpr IndexType NumRegs = OutputDimensions / OutputSimdWidth;
const auto input32 = reinterpret_cast<const std::int32_t*>(input);
const vec_t* biasvec = reinterpret_cast<const vec_t*>(biases);
vec_t acc[NumRegs];
for (IndexType k = 0; k < NumRegs; ++k)
acc[k] = biasvec[k];
for (IndexType i = 0; i < NumChunks; i += 2)
{
const vec_t in0 = vec_set_32(input32[i + 0]);
const vec_t in1 = vec_set_32(input32[i + 1]);
const auto col0 = reinterpret_cast<const vec_t*>(&weights[(i + 0) * OutputDimensions * 4]);
const auto col1 = reinterpret_cast<const vec_t*>(&weights[(i + 1) * OutputDimensions * 4]);
for (IndexType k = 0; k < NumRegs; ++k)
vec_add_dpbusd_32x2(acc[k], in0, col0[k], in1, col1[k]);
}
vec_t* outptr = reinterpret_cast<vec_t*>(output);
for (IndexType k = 0; k < NumRegs; ++k)
outptr[k] = acc[k];
}
else if constexpr (OutputDimensions == 1)
{
constexpr IndexType NumChunks = PaddedInputDimensions / SimdWidth;
vec_t sum0 = vec_setzero();
const auto row0 = reinterpret_cast<const vec_t*>(&weights[0]);
for (int j = 0; j < (int)NumChunks; ++j)
{
const vec_t in = inputVector[j];
vec_add_dpbusd_32(sum0, in, row0[j]);
}
output[0] = vec_hadd(sum0, biases[0]);
}
# undef vec_setzero
# undef vec_set_32
# undef vec_add_dpbusd_32
# undef vec_add_dpbusd_32x2
# undef vec_hadd
#else
// Use old implementation for the other architectures.
affine_transform_non_ssse3<
InputDimensions,
PaddedInputDimensions,
OutputDimensions>(output, weights, biases, input);
#endif
return output;
}
private:
using BiasType = OutputType;
using WeightType = std::int8_t;
alignas(CacheLineSize) BiasType biases[OutputDimensions];
alignas(CacheLineSize) WeightType weights[OutputDimensions * PaddedInputDimensions];
};
} // namespace Stockfish::Eval::NNUE::Layers
#endif // #ifndef NNUE_LAYERS_AFFINE_TRANSFORM_H_INCLUDED

286
src/Stockfish/src/nnue/layers/affine_transform_sparse_input.h Normal file
View file

@ -0,0 +1,286 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// Definition of layer AffineTransformSparseInput of NNUE evaluation function
#ifndef NNUE_LAYERS_AFFINE_TRANSFORM_SPARSE_INPUT_H_INCLUDED
#define NNUE_LAYERS_AFFINE_TRANSFORM_SPARSE_INPUT_H_INCLUDED
#include <iostream>
#include <algorithm>
#include <array>
#include <type_traits>
#include "../nnue_common.h"
#include "affine_transform.h"
#include "simd.h"
/*
This file contains the definition for a fully connected layer (aka affine transform) with block sparse input.
*/
namespace Stockfish::Eval::NNUE::Layers {
#if defined(__GNUC__) // GCC, Clang, ICC
static inline IndexType lsb_(std::uint32_t b) {
assert(b);
return IndexType(__builtin_ctzl(b));
}
#elif defined(_MSC_VER) // MSVC
static inline IndexType lsb_(std::uint32_t b) {
assert(b);
unsigned long idx;
_BitScanForward(&idx, b);
return (IndexType) idx;
}
#else // Compiler is neither GCC nor MSVC compatible
#error "Compiler not supported."
#endif
#if defined(USE_SSSE3)
alignas(CacheLineSize) static inline const std::array<std::array<std::uint16_t, 8>, 256> lookup_indices = [](){
std::array<std::array<std::uint16_t, 8>, 256> v{};
for (int i = 0; i < 256; ++i)
{
int j = i;
int k = 0;
while(j)
{
const IndexType lsbIndex = lsb_(std::uint32_t(j));
j &= j - 1;
v[i][k] = lsbIndex;
++k;
}
}
return v;
}();
alignas(CacheLineSize) static inline const std::array<unsigned, 256> lookup_count = [](){
std::array<unsigned, 256> v;
for (int i = 0; i < 256; ++i)
{
int j = i;
int k = 0;
while(j)
{
j &= j - 1;
++k;
}
v[i] = k;
}
return v;
}();
// Find indices of nonzero numbers in an int32_t array
template<const IndexType InputDimensions>
void find_nnz(const std::int32_t* input, std::uint16_t* out, IndexType& count_out) {
#if defined (USE_AVX512)
using vec_t = __m512i;
#define vec_nnz(a) _mm512_cmpgt_epi32_mask(a, _mm512_setzero_si512())
#elif defined (USE_AVX2)
using vec_t = __m256i;
#define vec_nnz(a) _mm256_movemask_ps(_mm256_castsi256_ps(_mm256_cmpgt_epi32(a, _mm256_setzero_si256())))
#elif defined (USE_SSSE3)
using vec_t = __m128i;
#define vec_nnz(a) _mm_movemask_ps(_mm_castsi128_ps(_mm_cmpgt_epi32(a, _mm_setzero_si128())))
#endif
constexpr IndexType InputSimdWidth = sizeof(vec_t) / sizeof(std::int32_t);
// Inputs are processed InputSimdWidth at a time and outputs are processed 8 at a time so we process in chunks of max(InputSimdWidth, 8)
constexpr IndexType ChunkSize = std::max<IndexType>(InputSimdWidth, 8);
constexpr IndexType NumChunks = InputDimensions / ChunkSize;
constexpr IndexType InputsPerChunk = ChunkSize / InputSimdWidth;
constexpr IndexType OutputsPerChunk = ChunkSize / 8;
const auto inputVector = reinterpret_cast<const vec_t*>(input);
IndexType count = 0;
__m128i base = _mm_set1_epi16(0);
__m128i increment = _mm_set1_epi16(8);
for (IndexType i = 0; i < NumChunks; ++i)
{
// bitmask of nonzero values in this chunk
unsigned nnz = 0;
for (IndexType j = 0; j < InputsPerChunk; ++j)
{
const vec_t inputChunk = inputVector[i * InputsPerChunk + j];
nnz |= (unsigned)vec_nnz(inputChunk) << (j * InputSimdWidth);
}
for (IndexType j = 0; j < OutputsPerChunk; ++j)
{
const auto lookup = (nnz >> (j * 8)) & 0xFF;
const auto offsets = _mm_loadu_si128(reinterpret_cast<const __m128i*>(&lookup_indices[lookup]));
_mm_storeu_si128(reinterpret_cast<__m128i*>(out + count), _mm_add_epi16(base, offsets));
count += lookup_count[lookup];
base = _mm_add_epi16(base, increment);
}
}
count_out = count;
}
# undef vec_nnz
#endif
// Sparse input implementation
template <IndexType InDims, IndexType OutDims>
class AffineTransformSparseInput {
public:
// Input/output type
// Input/output type
using InputType = std::uint8_t;
using OutputType = std::int32_t;
// Number of input/output dimensions
static constexpr IndexType InputDimensions = InDims;
static constexpr IndexType OutputDimensions = OutDims;
static_assert(OutputDimensions % 16 == 0, "Only implemented for OutputDimensions divisible by 16.");
static constexpr IndexType PaddedInputDimensions =
ceil_to_multiple<IndexType>(InputDimensions, MaxSimdWidth);
static constexpr IndexType PaddedOutputDimensions =
ceil_to_multiple<IndexType>(OutputDimensions, MaxSimdWidth);
#if defined (USE_SSSE3)
static constexpr IndexType ChunkSize = 4;
#else
static constexpr IndexType ChunkSize = 1;
#endif
using OutputBuffer = OutputType[PaddedOutputDimensions];
// Hash value embedded in the evaluation file
static constexpr std::uint32_t get_hash_value(std::uint32_t prevHash) {
std::uint32_t hashValue = 0xCC03DAE4u;
hashValue += OutputDimensions;
hashValue ^= prevHash >> 1;
hashValue ^= prevHash << 31;
return hashValue;
}
static IndexType get_weight_index_scrambled(IndexType i)
{
return
(i / ChunkSize) % (PaddedInputDimensions / ChunkSize) * OutputDimensions * ChunkSize +
i / PaddedInputDimensions * ChunkSize +
i % ChunkSize;
}
static IndexType get_weight_index(IndexType i)
{
#if defined (USE_SSSE3)
return get_weight_index_scrambled(i);
#else
return i;
#endif
}
// Read network parameters
bool read_parameters(std::istream& stream) {
read_little_endian<BiasType>(stream, biases, OutputDimensions);
for (IndexType i = 0; i < OutputDimensions * PaddedInputDimensions; ++i)
weights[get_weight_index(i)] = read_little_endian<WeightType>(stream);
return !stream.fail();
}
// Write network parameters
bool write_parameters(std::ostream& stream) const {
write_little_endian<BiasType>(stream, biases, OutputDimensions);
for (IndexType i = 0; i < OutputDimensions * PaddedInputDimensions; ++i)
write_little_endian<WeightType>(stream, weights[get_weight_index(i)]);
return !stream.fail();
}
// Forward propagation
const OutputType* propagate(
const InputType* input, OutputType* output) const {
#if defined (USE_SSSE3)
#if defined (USE_AVX512)
using vec_t = __m512i;
#define vec_setzero _mm512_setzero_si512
#define vec_set_32 _mm512_set1_epi32
#define vec_add_dpbusd_32 Simd::m512_add_dpbusd_epi32
#elif defined (USE_AVX2)
using vec_t = __m256i;
#define vec_setzero _mm256_setzero_si256
#define vec_set_32 _mm256_set1_epi32
#define vec_add_dpbusd_32 Simd::m256_add_dpbusd_epi32
#elif defined (USE_SSSE3)
using vec_t = __m128i;
#define vec_setzero _mm_setzero_si128
#define vec_set_32 _mm_set1_epi32
#define vec_add_dpbusd_32 Simd::m128_add_dpbusd_epi32
#endif
static constexpr IndexType OutputSimdWidth = sizeof(vec_t) / sizeof(OutputType);
constexpr IndexType NumChunks = ceil_to_multiple<IndexType>(InputDimensions, 8) / ChunkSize;
constexpr IndexType NumRegs = OutputDimensions / OutputSimdWidth;
std::uint16_t nnz[NumChunks];
IndexType count;
const auto input32 = reinterpret_cast<const std::int32_t*>(input);
// Find indices of nonzero 32bit blocks
find_nnz<NumChunks>(input32, nnz, count);
const vec_t* biasvec = reinterpret_cast<const vec_t*>(biases);
vec_t acc[NumRegs];
for (IndexType k = 0; k < NumRegs; ++k)
acc[k] = biasvec[k];
for (IndexType j = 0; j < count; ++j)
{
const auto i = nnz[j];
const vec_t in = vec_set_32(input32[i]);
const auto col = reinterpret_cast<const vec_t*>(&weights[i * OutputDimensions * ChunkSize]);
for (IndexType k = 0; k < NumRegs; ++k)
vec_add_dpbusd_32(acc[k], in, col[k]);
}
vec_t* outptr = reinterpret_cast<vec_t*>(output);
for (IndexType k = 0; k < NumRegs; ++k)
outptr[k] = acc[k];
# undef vec_setzero
# undef vec_set_32
# undef vec_add_dpbusd_32
#else
// Use dense implementation for the other architectures.
affine_transform_non_ssse3<
InputDimensions,
PaddedInputDimensions,
OutputDimensions>(output, weights, biases, input);
#endif
return output;
}
private:
using BiasType = OutputType;
using WeightType = std::int8_t;
alignas(CacheLineSize) BiasType biases[OutputDimensions];
alignas(CacheLineSize) WeightType weights[OutputDimensions * PaddedInputDimensions];
};
} // namespace Stockfish::Eval::NNUE::Layers
#endif // #ifndef NNUE_LAYERS_AFFINE_TRANSFORM_SPARSE_INPUT_H_INCLUDED

180
src/Stockfish/src/nnue/layers/clipped_relu.h Normal file
View file

@ -0,0 +1,180 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// Definition of layer ClippedReLU of NNUE evaluation function
#ifndef NNUE_LAYERS_CLIPPED_RELU_H_INCLUDED
#define NNUE_LAYERS_CLIPPED_RELU_H_INCLUDED
#include "../nnue_common.h"
namespace Stockfish::Eval::NNUE::Layers {
// Clipped ReLU
template <IndexType InDims>
class ClippedReLU {
public:
// Input/output type
using InputType = std::int32_t;
using OutputType = std::uint8_t;
// Number of input/output dimensions
static constexpr IndexType InputDimensions = InDims;
static constexpr IndexType OutputDimensions = InputDimensions;
static constexpr IndexType PaddedOutputDimensions =
ceil_to_multiple<IndexType>(OutputDimensions, 32);
using OutputBuffer = OutputType[PaddedOutputDimensions];
// Hash value embedded in the evaluation file
static constexpr std::uint32_t get_hash_value(std::uint32_t prevHash) {
std::uint32_t hashValue = 0x538D24C7u;
hashValue += prevHash;
return hashValue;
}
// Read network parameters
bool read_parameters(std::istream&) {
return true;
}
// Write network parameters
bool write_parameters(std::ostream&) const {
return true;
}
// Forward propagation
const OutputType* propagate(
const InputType* input, OutputType* output) const {
#if defined(USE_AVX2)
if constexpr (InputDimensions % SimdWidth == 0) {
constexpr IndexType NumChunks = InputDimensions / SimdWidth;
const __m256i Zero = _mm256_setzero_si256();
const __m256i Offsets = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0);
const auto in = reinterpret_cast<const __m256i*>(input);
const auto out = reinterpret_cast<__m256i*>(output);
for (IndexType i = 0; i < NumChunks; ++i) {
const __m256i words0 = _mm256_srai_epi16(_mm256_packs_epi32(
_mm256_load_si256(&in[i * 4 + 0]),
_mm256_load_si256(&in[i * 4 + 1])), WeightScaleBits);
const __m256i words1 = _mm256_srai_epi16(_mm256_packs_epi32(
_mm256_load_si256(&in[i * 4 + 2]),
_mm256_load_si256(&in[i * 4 + 3])), WeightScaleBits);
_mm256_store_si256(&out[i], _mm256_permutevar8x32_epi32(_mm256_max_epi8(
_mm256_packs_epi16(words0, words1), Zero), Offsets));
}
} else {
constexpr IndexType NumChunks = InputDimensions / (SimdWidth / 2);
const __m128i Zero = _mm_setzero_si128();
const auto in = reinterpret_cast<const __m128i*>(input);
const auto out = reinterpret_cast<__m128i*>(output);
for (IndexType i = 0; i < NumChunks; ++i) {
const __m128i words0 = _mm_srai_epi16(_mm_packs_epi32(
_mm_load_si128(&in[i * 4 + 0]),
_mm_load_si128(&in[i * 4 + 1])), WeightScaleBits);
const __m128i words1 = _mm_srai_epi16(_mm_packs_epi32(
_mm_load_si128(&in[i * 4 + 2]),
_mm_load_si128(&in[i * 4 + 3])), WeightScaleBits);
const __m128i packedbytes = _mm_packs_epi16(words0, words1);
_mm_store_si128(&out[i], _mm_max_epi8(packedbytes, Zero));
}
}
constexpr IndexType Start =
InputDimensions % SimdWidth == 0
? InputDimensions / SimdWidth * SimdWidth
: InputDimensions / (SimdWidth / 2) * (SimdWidth / 2);
#elif defined(USE_SSE2)
constexpr IndexType NumChunks = InputDimensions / SimdWidth;
#ifdef USE_SSE41
const __m128i Zero = _mm_setzero_si128();
#else
const __m128i k0x80s = _mm_set1_epi8(-128);
#endif
const auto in = reinterpret_cast<const __m128i*>(input);
const auto out = reinterpret_cast<__m128i*>(output);
for (IndexType i = 0; i < NumChunks; ++i) {
const __m128i words0 = _mm_srai_epi16(_mm_packs_epi32(
_mm_load_si128(&in[i * 4 + 0]),
_mm_load_si128(&in[i * 4 + 1])), WeightScaleBits);
const __m128i words1 = _mm_srai_epi16(_mm_packs_epi32(
_mm_load_si128(&in[i * 4 + 2]),
_mm_load_si128(&in[i * 4 + 3])), WeightScaleBits);
const __m128i packedbytes = _mm_packs_epi16(words0, words1);
_mm_store_si128(&out[i],
#ifdef USE_SSE41
_mm_max_epi8(packedbytes, Zero)
#else
_mm_subs_epi8(_mm_adds_epi8(packedbytes, k0x80s), k0x80s)
#endif
);
}
constexpr IndexType Start = NumChunks * SimdWidth;
#elif defined(USE_MMX)
constexpr IndexType NumChunks = InputDimensions / SimdWidth;
const __m64 k0x80s = _mm_set1_pi8(-128);
const auto in = reinterpret_cast<const __m64*>(input);
const auto out = reinterpret_cast<__m64*>(output);
for (IndexType i = 0; i < NumChunks; ++i) {
const __m64 words0 = _mm_srai_pi16(
_mm_packs_pi32(in[i * 4 + 0], in[i * 4 + 1]),
WeightScaleBits);
const __m64 words1 = _mm_srai_pi16(
_mm_packs_pi32(in[i * 4 + 2], in[i * 4 + 3]),
WeightScaleBits);
const __m64 packedbytes = _mm_packs_pi16(words0, words1);
out[i] = _mm_subs_pi8(_mm_adds_pi8(packedbytes, k0x80s), k0x80s);
}
_mm_empty();
constexpr IndexType Start = NumChunks * SimdWidth;
#elif defined(USE_NEON)
constexpr IndexType NumChunks = InputDimensions / (SimdWidth / 2);
const int8x8_t Zero = {0};
const auto in = reinterpret_cast<const int32x4_t*>(input);
const auto out = reinterpret_cast<int8x8_t*>(output);
for (IndexType i = 0; i < NumChunks; ++i) {
int16x8_t shifted;
const auto pack = reinterpret_cast<int16x4_t*>(&shifted);
pack[0] = vqshrn_n_s32(in[i * 2 + 0], WeightScaleBits);
pack[1] = vqshrn_n_s32(in[i * 2 + 1], WeightScaleBits);
out[i] = vmax_s8(vqmovn_s16(shifted), Zero);
}
constexpr IndexType Start = NumChunks * (SimdWidth / 2);
#else
constexpr IndexType Start = 0;
#endif
for (IndexType i = Start; i < InputDimensions; ++i) {
output[i] = static_cast<OutputType>(
std::max(0, std::min(127, input[i] >> WeightScaleBits)));
}
return output;
}
};
} // namespace Stockfish::Eval::NNUE::Layers
#endif // NNUE_LAYERS_CLIPPED_RELU_H_INCLUDED

403
src/Stockfish/src/nnue/layers/simd.h Normal file
View file

@ -0,0 +1,403 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef STOCKFISH_SIMD_H_INCLUDED
#define STOCKFISH_SIMD_H_INCLUDED
#if defined(USE_AVX2)
# include <immintrin.h>
#elif defined(USE_SSE41)
# include <smmintrin.h>
#elif defined(USE_SSSE3)
# include <tmmintrin.h>
#elif defined(USE_SSE2)
# include <emmintrin.h>
#elif defined(USE_MMX)
# include <mmintrin.h>
#elif defined(USE_NEON)
# include <arm_neon.h>
#endif
// The inline asm is only safe for GCC, where it is necessary to get good codegen.
// See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101693
// Clang does fine without it.
// Play around here: https://godbolt.org/z/7EWqrYq51
#if (defined(__GNUC__) && !defined(__clang__) && !defined(__INTEL_COMPILER))
#define USE_INLINE_ASM
#endif
// Use either the AVX512 or AVX-VNNI version of the VNNI instructions.
#if defined(USE_AVXVNNI)
#define VNNI_PREFIX "%{vex%} "
#else
#define VNNI_PREFIX ""
#endif
namespace Stockfish::Simd {
#if defined (USE_AVX512)
[[maybe_unused]] static int m512_hadd(__m512i sum, int bias) {
return _mm512_reduce_add_epi32(sum) + bias;
}
/*
Parameters:
sum0 = [zmm0.i128[0], zmm0.i128[1], zmm0.i128[2], zmm0.i128[3]]
sum1 = [zmm1.i128[0], zmm1.i128[1], zmm1.i128[2], zmm1.i128[3]]
sum2 = [zmm2.i128[0], zmm2.i128[1], zmm2.i128[2], zmm2.i128[3]]
sum3 = [zmm3.i128[0], zmm3.i128[1], zmm3.i128[2], zmm3.i128[3]]
Returns:
ret = [
reduce_add_epi32(zmm0.i128[0]), reduce_add_epi32(zmm1.i128[0]), reduce_add_epi32(zmm2.i128[0]), reduce_add_epi32(zmm3.i128[0]),
reduce_add_epi32(zmm0.i128[1]), reduce_add_epi32(zmm1.i128[1]), reduce_add_epi32(zmm2.i128[1]), reduce_add_epi32(zmm3.i128[1]),
reduce_add_epi32(zmm0.i128[2]), reduce_add_epi32(zmm1.i128[2]), reduce_add_epi32(zmm2.i128[2]), reduce_add_epi32(zmm3.i128[2]),
reduce_add_epi32(zmm0.i128[3]), reduce_add_epi32(zmm1.i128[3]), reduce_add_epi32(zmm2.i128[3]), reduce_add_epi32(zmm3.i128[3])
]
*/
[[maybe_unused]] static __m512i m512_hadd128x16_interleave(
__m512i sum0, __m512i sum1, __m512i sum2, __m512i sum3) {
__m512i sum01a = _mm512_unpacklo_epi32(sum0, sum1);
__m512i sum01b = _mm512_unpackhi_epi32(sum0, sum1);
__m512i sum23a = _mm512_unpacklo_epi32(sum2, sum3);
__m512i sum23b = _mm512_unpackhi_epi32(sum2, sum3);
__m512i sum01 = _mm512_add_epi32(sum01a, sum01b);
__m512i sum23 = _mm512_add_epi32(sum23a, sum23b);
__m512i sum0123a = _mm512_unpacklo_epi64(sum01, sum23);
__m512i sum0123b = _mm512_unpackhi_epi64(sum01, sum23);
return _mm512_add_epi32(sum0123a, sum0123b);
}
[[maybe_unused]] static __m128i m512_haddx4(
__m512i sum0, __m512i sum1, __m512i sum2, __m512i sum3,
__m128i bias) {
__m512i sum = m512_hadd128x16_interleave(sum0, sum1, sum2, sum3);
__m256i sum256lo = _mm512_castsi512_si256(sum);
__m256i sum256hi = _mm512_extracti64x4_epi64(sum, 1);
sum256lo = _mm256_add_epi32(sum256lo, sum256hi);
__m128i sum128lo = _mm256_castsi256_si128(sum256lo);
__m128i sum128hi = _mm256_extracti128_si256(sum256lo, 1);
return _mm_add_epi32(_mm_add_epi32(sum128lo, sum128hi), bias);
}
[[maybe_unused]] static void m512_add_dpbusd_epi32(
__m512i& acc,
__m512i a,
__m512i b) {
# if defined (USE_VNNI)
# if defined (USE_INLINE_ASM)
asm(
"vpdpbusd %[b], %[a], %[acc]\n\t"
: [acc]"+v"(acc)
: [a]"v"(a), [b]"vm"(b)
);
# else
acc = _mm512_dpbusd_epi32(acc, a, b);
# endif
# else
# if defined (USE_INLINE_ASM)
__m512i tmp = _mm512_maddubs_epi16(a, b);
asm(
"vpmaddwd %[tmp], %[ones], %[tmp]\n\t"
"vpaddd %[acc], %[tmp], %[acc]\n\t"
: [acc]"+v"(acc), [tmp]"+&v"(tmp)
: [ones]"v"(_mm512_set1_epi16(1))
);
# else
__m512i product0 = _mm512_maddubs_epi16(a, b);
product0 = _mm512_madd_epi16(product0, _mm512_set1_epi16(1));
acc = _mm512_add_epi32(acc, product0);
# endif
# endif
}
[[maybe_unused]] static void m512_add_dpbusd_epi32x2(
__m512i& acc,
__m512i a0, __m512i b0,
__m512i a1, __m512i b1) {
# if defined (USE_VNNI)
# if defined (USE_INLINE_ASM)
asm(
"vpdpbusd %[b0], %[a0], %[acc]\n\t"
"vpdpbusd %[b1], %[a1], %[acc]\n\t"
: [acc]"+&v"(acc)
: [a0]"v"(a0), [b0]"vm"(b0), [a1]"v"(a1), [b1]"vm"(b1)
);
# else
acc = _mm512_dpbusd_epi32(acc, a0, b0);
acc = _mm512_dpbusd_epi32(acc, a1, b1);
# endif
# else
# if defined (USE_INLINE_ASM)
__m512i tmp0 = _mm512_maddubs_epi16(a0, b0);
__m512i tmp1 = _mm512_maddubs_epi16(a1, b1);
asm(
"vpmaddwd %[tmp0], %[ones], %[tmp0]\n\t"
"vpmaddwd %[tmp1], %[ones], %[tmp1]\n\t"
"vpaddd %[tmp0], %[tmp1], %[tmp0]\n\t"
"vpaddd %[acc], %[tmp0], %[acc]\n\t"
: [acc]"+v"(acc), [tmp0]"+&v"(tmp0), [tmp1]"+&v"(tmp1)
: [ones]"v"(_mm512_set1_epi16(1))
);
# else
__m512i product0 = _mm512_maddubs_epi16(a0, b0);
__m512i product1 = _mm512_maddubs_epi16(a1, b1);
product0 = _mm512_madd_epi16(product0, _mm512_set1_epi16(1));
product1 = _mm512_madd_epi16(product1, _mm512_set1_epi16(1));
acc = _mm512_add_epi32(acc, _mm512_add_epi32(product0, product1));
# endif
# endif
}
#endif
#if defined (USE_AVX2)
[[maybe_unused]] static int m256_hadd(__m256i sum, int bias) {
__m128i sum128 = _mm_add_epi32(_mm256_castsi256_si128(sum), _mm256_extracti128_si256(sum, 1));
sum128 = _mm_add_epi32(sum128, _mm_shuffle_epi32(sum128, _MM_PERM_BADC));
sum128 = _mm_add_epi32(sum128, _mm_shuffle_epi32(sum128, _MM_PERM_CDAB));
return _mm_cvtsi128_si32(sum128) + bias;
}
[[maybe_unused]] static __m128i m256_haddx4(
__m256i sum0, __m256i sum1, __m256i sum2, __m256i sum3,
__m128i bias) {
sum0 = _mm256_hadd_epi32(sum0, sum1);
sum2 = _mm256_hadd_epi32(sum2, sum3);
sum0 = _mm256_hadd_epi32(sum0, sum2);
__m128i sum128lo = _mm256_castsi256_si128(sum0);
__m128i sum128hi = _mm256_extracti128_si256(sum0, 1);
return _mm_add_epi32(_mm_add_epi32(sum128lo, sum128hi), bias);
}
[[maybe_unused]] static void m256_add_dpbusd_epi32(
__m256i& acc,
__m256i a,
__m256i b) {
# if defined (USE_VNNI)
# if defined (USE_INLINE_ASM)
asm(
VNNI_PREFIX "vpdpbusd %[b], %[a], %[acc]\n\t"
: [acc]"+v"(acc)
: [a]"v"(a), [b]"vm"(b)
);
# else
acc = _mm256_dpbusd_epi32(acc, a, b);
# endif
# else
# if defined (USE_INLINE_ASM)
__m256i tmp = _mm256_maddubs_epi16(a, b);
asm(
"vpmaddwd %[tmp], %[ones], %[tmp]\n\t"
"vpaddd %[acc], %[tmp], %[acc]\n\t"
: [acc]"+v"(acc), [tmp]"+&v"(tmp)
: [ones]"v"(_mm256_set1_epi16(1))
);
# else
__m256i product0 = _mm256_maddubs_epi16(a, b);
product0 = _mm256_madd_epi16(product0, _mm256_set1_epi16(1));
acc = _mm256_add_epi32(acc, product0);
# endif
# endif
}
[[maybe_unused]] static void m256_add_dpbusd_epi32x2(
__m256i& acc,
__m256i a0, __m256i b0,
__m256i a1, __m256i b1) {
# if defined (USE_VNNI)
# if defined (USE_INLINE_ASM)
asm(
VNNI_PREFIX "vpdpbusd %[b0], %[a0], %[acc]\n\t"
VNNI_PREFIX "vpdpbusd %[b1], %[a1], %[acc]\n\t"
: [acc]"+&v"(acc)
: [a0]"v"(a0), [b0]"vm"(b0), [a1]"v"(a1), [b1]"vm"(b1)
);
# else
acc = _mm256_dpbusd_epi32(acc, a0, b0);
acc = _mm256_dpbusd_epi32(acc, a1, b1);
# endif
# else
# if defined (USE_INLINE_ASM)
__m256i tmp0 = _mm256_maddubs_epi16(a0, b0);
__m256i tmp1 = _mm256_maddubs_epi16(a1, b1);
asm(
"vpmaddwd %[tmp0], %[ones], %[tmp0]\n\t"
"vpmaddwd %[tmp1], %[ones], %[tmp1]\n\t"
"vpaddd %[tmp0], %[tmp1], %[tmp0]\n\t"
"vpaddd %[acc], %[tmp0], %[acc]\n\t"
: [acc]"+v"(acc), [tmp0]"+&v"(tmp0), [tmp1]"+&v"(tmp1)
: [ones]"v"(_mm256_set1_epi16(1))
);
# else
__m256i product0 = _mm256_maddubs_epi16(a0, b0);
__m256i product1 = _mm256_maddubs_epi16(a1, b1);
product0 = _mm256_madd_epi16(product0, _mm256_set1_epi16(1));
product1 = _mm256_madd_epi16(product1, _mm256_set1_epi16(1));
acc = _mm256_add_epi32(acc, _mm256_add_epi32(product0, product1));
# endif
# endif
}
#endif
#if defined (USE_SSSE3)
[[maybe_unused]] static int m128_hadd(__m128i sum, int bias) {
sum = _mm_add_epi32(sum, _mm_shuffle_epi32(sum, 0x4E)); //_MM_PERM_BADC
sum = _mm_add_epi32(sum, _mm_shuffle_epi32(sum, 0xB1)); //_MM_PERM_CDAB
return _mm_cvtsi128_si32(sum) + bias;
}
[[maybe_unused]] static __m128i m128_haddx4(
__m128i sum0, __m128i sum1, __m128i sum2, __m128i sum3,
__m128i bias) {
sum0 = _mm_hadd_epi32(sum0, sum1);
sum2 = _mm_hadd_epi32(sum2, sum3);
sum0 = _mm_hadd_epi32(sum0, sum2);
return _mm_add_epi32(sum0, bias);
}
[[maybe_unused]] static void m128_add_dpbusd_epi32(
__m128i& acc,
__m128i a,
__m128i b) {
# if defined (USE_INLINE_ASM)
__m128i tmp = _mm_maddubs_epi16(a, b);
asm(
"pmaddwd %[ones], %[tmp]\n\t"
"paddd %[tmp], %[acc]\n\t"
: [acc]"+v"(acc), [tmp]"+&v"(tmp)
: [ones]"v"(_mm_set1_epi16(1))
);
# else
__m128i product0 = _mm_maddubs_epi16(a, b);
product0 = _mm_madd_epi16(product0, _mm_set1_epi16(1));
acc = _mm_add_epi32(acc, product0);
# endif
}
[[maybe_unused]] static void m128_add_dpbusd_epi32x2(
__m128i& acc,
__m128i a0, __m128i b0,
__m128i a1, __m128i b1) {
# if defined (USE_INLINE_ASM)
__m128i tmp0 = _mm_maddubs_epi16(a0, b0);
__m128i tmp1 = _mm_maddubs_epi16(a1, b1);
asm(
"pmaddwd %[ones], %[tmp0]\n\t"
"pmaddwd %[ones], %[tmp1]\n\t"
"paddd %[tmp1], %[tmp0]\n\t"
"paddd %[tmp0], %[acc]\n\t"
: [acc]"+v"(acc), [tmp0]"+&v"(tmp0), [tmp1]"+&v"(tmp1)
: [ones]"v"(_mm_set1_epi16(1))
);
# else
__m128i product0 = _mm_maddubs_epi16(a0, b0);
__m128i product1 = _mm_maddubs_epi16(a1, b1);
product0 = _mm_madd_epi16(product0, _mm_set1_epi16(1));
product1 = _mm_madd_epi16(product1, _mm_set1_epi16(1));
acc = _mm_add_epi32(acc, _mm_add_epi32(product0, product1));
# endif
}
#endif
#if defined (USE_NEON_DOTPROD)
[[maybe_unused]] static void dotprod_m128_add_dpbusd_epi32x2(
int32x4_t& acc,
int8x16_t a0, int8x16_t b0,
int8x16_t a1, int8x16_t b1) {
acc = vdotq_s32(acc, a0, b0);
acc = vdotq_s32(acc, a1, b1);
}
#endif
#if defined (USE_NEON)
[[maybe_unused]] static int neon_m128_reduce_add_epi32(int32x4_t s) {
# if USE_NEON >= 8
return vaddvq_s32(s);
# else
return s[0] + s[1] + s[2] + s[3];
# endif
}
[[maybe_unused]] static int neon_m128_hadd(int32x4_t sum, int bias) {
return neon_m128_reduce_add_epi32(sum) + bias;
}
[[maybe_unused]] static int32x4_t neon_m128_haddx4(
int32x4_t sum0, int32x4_t sum1, int32x4_t sum2, int32x4_t sum3,
int32x4_t bias) {
int32x4_t hsums {
neon_m128_reduce_add_epi32(sum0),
neon_m128_reduce_add_epi32(sum1),
neon_m128_reduce_add_epi32(sum2),
neon_m128_reduce_add_epi32(sum3)
};
return vaddq_s32(hsums, bias);
}
[[maybe_unused]] static void neon_m128_add_dpbusd_epi32x2(
int32x4_t& acc,
int8x8_t a0, int8x8_t b0,
int8x8_t a1, int8x8_t b1) {
int16x8_t product = vmull_s8(a0, b0);
product = vmlal_s8(product, a1, b1);
acc = vpadalq_s16(acc, product);
}
#endif
}
#endif // STOCKFISH_SIMD_H_INCLUDED

120
src/Stockfish/src/nnue/layers/sqr_clipped_relu.h Normal file
View file

@ -0,0 +1,120 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// Definition of layer ClippedReLU of NNUE evaluation function
#ifndef NNUE_LAYERS_SQR_CLIPPED_RELU_H_INCLUDED
#define NNUE_LAYERS_SQR_CLIPPED_RELU_H_INCLUDED
#include "../nnue_common.h"
namespace Stockfish::Eval::NNUE::Layers {
// Clipped ReLU
template <IndexType InDims>
class SqrClippedReLU {
public:
// Input/output type
using InputType = std::int32_t;
using OutputType = std::uint8_t;
// Number of input/output dimensions
static constexpr IndexType InputDimensions = InDims;
static constexpr IndexType OutputDimensions = InputDimensions;
static constexpr IndexType PaddedOutputDimensions =
ceil_to_multiple<IndexType>(OutputDimensions, 32);
using OutputBuffer = OutputType[PaddedOutputDimensions];
// Hash value embedded in the evaluation file
static constexpr std::uint32_t get_hash_value(std::uint32_t prevHash) {
std::uint32_t hashValue = 0x538D24C7u;
hashValue += prevHash;
return hashValue;
}
// Read network parameters
bool read_parameters(std::istream&) {
return true;
}
// Write network parameters
bool write_parameters(std::ostream&) const {
return true;
}
// Forward propagation
const OutputType* propagate(
const InputType* input, OutputType* output) const {
#if defined(USE_SSE2)
constexpr IndexType NumChunks = InputDimensions / 16;
#ifdef USE_SSE41
const __m128i Zero = _mm_setzero_si128();
#else
const __m128i k0x80s = _mm_set1_epi8(-128);
#endif
static_assert(WeightScaleBits == 6);
const auto in = reinterpret_cast<const __m128i*>(input);
const auto out = reinterpret_cast<__m128i*>(output);
for (IndexType i = 0; i < NumChunks; ++i) {
__m128i words0 = _mm_packs_epi32(
_mm_load_si128(&in[i * 4 + 0]),
_mm_load_si128(&in[i * 4 + 1]));
__m128i words1 = _mm_packs_epi32(
_mm_load_si128(&in[i * 4 + 2]),
_mm_load_si128(&in[i * 4 + 3]));
// Not sure if
words0 = _mm_srli_epi16(_mm_mulhi_epi16(words0, words0), 3);
words1 = _mm_srli_epi16(_mm_mulhi_epi16(words1, words1), 3);
const __m128i packedbytes = _mm_packs_epi16(words0, words1);
_mm_store_si128(&out[i],
#ifdef USE_SSE41
_mm_max_epi8(packedbytes, Zero)
#else
_mm_subs_epi8(_mm_adds_epi8(packedbytes, k0x80s), k0x80s)
#endif
);
}
constexpr IndexType Start = NumChunks * 16;
#else
constexpr IndexType Start = 0;
#endif
for (IndexType i = Start; i < InputDimensions; ++i) {
output[i] = static_cast<OutputType>(
// really should be /127 but we need to make it fast
// needs to be accounted for in the trainer
std::max(0ll, std::min(127ll, (((long long)input[i] * input[i]) >> (2 * WeightScaleBits)) / 128)));
}
return output;
}
};
} // namespace Stockfish::Eval::NNUE::Layers
#endif // NNUE_LAYERS_SQR_CLIPPED_RELU_H_INCLUDED

37
src/Stockfish/src/nnue/nnue_accumulator.h Normal file
View file

@ -0,0 +1,37 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// Class for difference calculation of NNUE evaluation function
#ifndef NNUE_ACCUMULATOR_H_INCLUDED
#define NNUE_ACCUMULATOR_H_INCLUDED
#include "nnue_architecture.h"
namespace Stockfish::Eval::NNUE {
// Class that holds the result of affine transformation of input features
struct alignas(CacheLineSize) Accumulator {
std::int16_t accumulation[2][TransformedFeatureDimensions];
std::int32_t psqtAccumulation[2][PSQTBuckets];
bool computed[2];
};
} // namespace Stockfish::Eval::NNUE
#endif // NNUE_ACCUMULATOR_H_INCLUDED

137
src/Stockfish/src/nnue/nnue_architecture.h Normal file
View file

@ -0,0 +1,137 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// Input features and network structure used in NNUE evaluation function
#ifndef NNUE_ARCHITECTURE_H_INCLUDED
#define NNUE_ARCHITECTURE_H_INCLUDED
#include <memory>
#include "nnue_common.h"
#include "features/half_ka_v2_hm.h"
#include "layers/affine_transform_sparse_input.h"
#include "layers/affine_transform.h"
#include "layers/clipped_relu.h"
#include "layers/sqr_clipped_relu.h"
#include "../misc.h"
namespace Stockfish::Eval::NNUE {
// Input features used in evaluation function
using FeatureSet = Features::HalfKAv2_hm;
// Number of input feature dimensions after conversion
constexpr IndexType TransformedFeatureDimensions = 1536;
constexpr IndexType PSQTBuckets = 8;
constexpr IndexType LayerStacks = 8;
struct Network
{
static constexpr int FC_0_OUTPUTS = 15;
static constexpr int FC_1_OUTPUTS = 32;
Layers::AffineTransformSparseInput<TransformedFeatureDimensions, FC_0_OUTPUTS + 1> fc_0;
Layers::SqrClippedReLU<FC_0_OUTPUTS + 1> ac_sqr_0;
Layers::ClippedReLU<FC_0_OUTPUTS + 1> ac_0;
Layers::AffineTransform<FC_0_OUTPUTS * 2, FC_1_OUTPUTS> fc_1;
Layers::ClippedReLU<FC_1_OUTPUTS> ac_1;
Layers::AffineTransform<FC_1_OUTPUTS, 1> fc_2;
// Hash value embedded in the evaluation file
static constexpr std::uint32_t get_hash_value() {
// input slice hash
std::uint32_t hashValue = 0xEC42E90Du;
hashValue ^= TransformedFeatureDimensions * 2;
hashValue = decltype(fc_0)::get_hash_value(hashValue);
hashValue = decltype(ac_0)::get_hash_value(hashValue);
hashValue = decltype(fc_1)::get_hash_value(hashValue);
hashValue = decltype(ac_1)::get_hash_value(hashValue);
hashValue = decltype(fc_2)::get_hash_value(hashValue);
return hashValue;
}
// Read network parameters
bool read_parameters(std::istream& stream) {
return fc_0.read_parameters(stream)
&& ac_0.read_parameters(stream)
&& fc_1.read_parameters(stream)
&& ac_1.read_parameters(stream)
&& fc_2.read_parameters(stream);
}
// Write network parameters
bool write_parameters(std::ostream& stream) const {
return fc_0.write_parameters(stream)
&& ac_0.write_parameters(stream)
&& fc_1.write_parameters(stream)
&& ac_1.write_parameters(stream)
&& fc_2.write_parameters(stream);
}
std::int32_t propagate(const TransformedFeatureType* transformedFeatures)
{
struct alignas(CacheLineSize) Buffer
{
alignas(CacheLineSize) decltype(fc_0)::OutputBuffer fc_0_out;
alignas(CacheLineSize) decltype(ac_sqr_0)::OutputType ac_sqr_0_out[ceil_to_multiple<IndexType>(FC_0_OUTPUTS * 2, 32)];
alignas(CacheLineSize) decltype(ac_0)::OutputBuffer ac_0_out;
alignas(CacheLineSize) decltype(fc_1)::OutputBuffer fc_1_out;
alignas(CacheLineSize) decltype(ac_1)::OutputBuffer ac_1_out;
alignas(CacheLineSize) decltype(fc_2)::OutputBuffer fc_2_out;
Buffer()
{
std::memset(this, 0, sizeof(*this));
}
};
#if defined(__clang__) && (__APPLE__)
// workaround for a bug reported with xcode 12
static thread_local auto tlsBuffer = std::make_unique<Buffer>();
// Access TLS only once, cache result.
Buffer& buffer = *tlsBuffer;
#else
alignas(CacheLineSize) static thread_local Buffer buffer;
#endif
fc_0.propagate(transformedFeatures, buffer.fc_0_out);
ac_sqr_0.propagate(buffer.fc_0_out, buffer.ac_sqr_0_out);
ac_0.propagate(buffer.fc_0_out, buffer.ac_0_out);
std::memcpy(buffer.ac_sqr_0_out + FC_0_OUTPUTS, buffer.ac_0_out, FC_0_OUTPUTS * sizeof(decltype(ac_0)::OutputType));
fc_1.propagate(buffer.ac_sqr_0_out, buffer.fc_1_out);
ac_1.propagate(buffer.fc_1_out, buffer.ac_1_out);
fc_2.propagate(buffer.ac_1_out, buffer.fc_2_out);
// buffer.fc_0_out[FC_0_OUTPUTS] is such that 1.0 is equal to 127*(1<<WeightScaleBits) in quantized form
// but we want 1.0 to be equal to 600*OutputScale
std::int32_t fwdOut = int(buffer.fc_0_out[FC_0_OUTPUTS]) * (600*OutputScale) / (127*(1<<WeightScaleBits));
std::int32_t outputValue = buffer.fc_2_out[0] + fwdOut;
return outputValue;
}
};
} // namespace Stockfish::Eval::NNUE
#endif // #ifndef NNUE_ARCHITECTURE_H_INCLUDED

241
src/Stockfish/src/nnue/nnue_common.h Normal file
View file

@ -0,0 +1,241 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// Constants used in NNUE evaluation function
#ifndef NNUE_COMMON_H_INCLUDED
#define NNUE_COMMON_H_INCLUDED
#include <cstring>
#include <iostream>
#include "../misc.h" // for IsLittleEndian
#if defined(USE_AVX2)
#include <immintrin.h>
#elif defined(USE_SSE41)
#include <smmintrin.h>
#elif defined(USE_SSSE3)
#include <tmmintrin.h>
#elif defined(USE_SSE2)
#include <emmintrin.h>
#elif defined(USE_MMX)
#include <mmintrin.h>
#elif defined(USE_NEON)
#include <arm_neon.h>
#endif
namespace Stockfish::Eval::NNUE {
// Version of the evaluation file
constexpr std::uint32_t Version = 0x7AF32F20u;
// Constant used in evaluation value calculation
constexpr int OutputScale = 16;
constexpr int WeightScaleBits = 6;
// Size of cache line (in bytes)
constexpr std::size_t CacheLineSize = 64;
constexpr const char Leb128MagicString[] = "COMPRESSED_LEB128";
constexpr const std::size_t Leb128MagicStringSize = sizeof(Leb128MagicString) - 1;
// SIMD width (in bytes)
#if defined(USE_AVX2)
constexpr std::size_t SimdWidth = 32;
#elif defined(USE_SSE2)
constexpr std::size_t SimdWidth = 16;
#elif defined(USE_MMX)
constexpr std::size_t SimdWidth = 8;
#elif defined(USE_NEON)
constexpr std::size_t SimdWidth = 16;
#endif
constexpr std::size_t MaxSimdWidth = 32;
// Type of input feature after conversion
using TransformedFeatureType = std::uint8_t;
using IndexType = std::uint32_t;
// Round n up to be a multiple of base
template <typename IntType>
constexpr IntType ceil_to_multiple(IntType n, IntType base) {
return (n + base - 1) / base * base;
}
// read_little_endian() is our utility to read an integer (signed or unsigned, any size)
// from a stream in little-endian order. We swap the byte order after the read if
// necessary to return a result with the byte ordering of the compiling machine.
template <typename IntType>
inline IntType read_little_endian(std::istream& stream) {
IntType result;
if (IsLittleEndian)
stream.read(reinterpret_cast<char*>(&result), sizeof(IntType));
else
{
std::uint8_t u[sizeof(IntType)];
typename std::make_unsigned<IntType>::type v = 0;
stream.read(reinterpret_cast<char*>(u), sizeof(IntType));
for (std::size_t i = 0; i < sizeof(IntType); ++i)
v = (v << 8) | u[sizeof(IntType) - i - 1];
std::memcpy(&result, &v, sizeof(IntType));
}
return result;
}
// write_little_endian() is our utility to write an integer (signed or unsigned, any size)
// to a stream in little-endian order. We swap the byte order before the write if
// necessary to always write in little endian order, independently of the byte
// ordering of the compiling machine.
template <typename IntType>
inline void write_little_endian(std::ostream& stream, IntType value) {
if (IsLittleEndian)
stream.write(reinterpret_cast<const char*>(&value), sizeof(IntType));
else
{
std::uint8_t u[sizeof(IntType)];
typename std::make_unsigned<IntType>::type v = value;
std::size_t i = 0;
// if constexpr to silence the warning about shift by 8
if constexpr (sizeof(IntType) > 1)
{
for (; i + 1 < sizeof(IntType); ++i)
{
u[i] = (std::uint8_t)v;
v >>= 8;
}
}
u[i] = (std::uint8_t)v;
stream.write(reinterpret_cast<char*>(u), sizeof(IntType));
}
}
// read_little_endian(s, out, N) : read integers in bulk from a little indian stream.
// This reads N integers from stream s and put them in array out.
template <typename IntType>
inline void read_little_endian(std::istream& stream, IntType* out, std::size_t count) {
if (IsLittleEndian)
stream.read(reinterpret_cast<char*>(out), sizeof(IntType) * count);
else
for (std::size_t i = 0; i < count; ++i)
out[i] = read_little_endian<IntType>(stream);
}
// write_little_endian(s, values, N) : write integers in bulk to a little indian stream.
// This takes N integers from array values and writes them on stream s.
template <typename IntType>
inline void write_little_endian(std::ostream& stream, const IntType* values, std::size_t count) {
if (IsLittleEndian)
stream.write(reinterpret_cast<const char*>(values), sizeof(IntType) * count);
else
for (std::size_t i = 0; i < count; ++i)
write_little_endian<IntType>(stream, values[i]);
}
template <typename IntType>
inline void read_leb_128(std::istream& stream, IntType* out, std::size_t count) {
static_assert(std::is_signed_v<IntType>, "Not implemented for unsigned types");
char leb128MagicString[Leb128MagicStringSize];
stream.read(leb128MagicString, Leb128MagicStringSize);
assert(strncmp(Leb128MagicString, leb128MagicString, Leb128MagicStringSize) == 0);
const std::uint32_t BUF_SIZE = 4096;
std::uint8_t buf[BUF_SIZE];
auto bytes_left = read_little_endian<std::uint32_t>(stream);
std::uint32_t buf_pos = BUF_SIZE;
for (std::size_t i = 0; i < count; ++i) {
IntType result = 0;
size_t shift = 0;
do {
if (buf_pos == BUF_SIZE) {
stream.read(reinterpret_cast<char*>(buf), std::min(bytes_left, BUF_SIZE));
buf_pos = 0;
}
std::uint8_t byte = buf[buf_pos++];
--bytes_left;
result |= (byte & 0x7f) << shift;
shift += 7;
if ((byte & 0x80) == 0) {
out[i] = sizeof(IntType) * 8 <= shift || (byte & 0x40) == 0 ? result : result | ~((1 << shift) - 1);
break;
}
} while (shift < sizeof(IntType) * 8);
}
assert(bytes_left == 0);
}
template <typename IntType>
inline void write_leb_128(std::ostream& stream, const IntType* values, std::size_t count) {
static_assert(std::is_signed_v<IntType>, "Not implemented for unsigned types");
stream.write(Leb128MagicString, Leb128MagicStringSize);
std::uint32_t byte_count = 0;
for (std::size_t i = 0; i < count; ++i) {
IntType value = values[i];
std::uint8_t byte;
do {
byte = value & 0x7f;
value >>= 7;
++byte_count;
} while ((byte & 0x40) == 0 ? value != 0 : value != -1);
}
write_little_endian(stream, byte_count);
const std::uint32_t BUF_SIZE = 4096;
std::uint8_t buf[BUF_SIZE];
std::uint32_t buf_pos = 0;
auto flush = [&]() {
if (buf_pos > 0) {
stream.write(reinterpret_cast<char*>(buf), buf_pos);
buf_pos = 0;
}
};
auto write = [&](std::uint8_t byte) {
buf[buf_pos++] = byte;
if (buf_pos == BUF_SIZE) flush();
};
for (std::size_t i = 0; i < count; ++i) {
IntType value = values[i];
while (true) {
std::uint8_t byte = value & 0x7f;
value >>= 7;
if ((byte & 0x40) == 0 ? value == 0 : value == -1) {
write(byte);
break;
}
write(byte | 0x80);
}
}
flush();
}
} // namespace Stockfish::Eval::NNUE
#endif // #ifndef NNUE_COMMON_H_INCLUDED

674
src/Stockfish/src/nnue/nnue_feature_transformer.h Normal file
View file

@ -0,0 +1,674 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// A class that converts the input features of the NNUE evaluation function
#ifndef NNUE_FEATURE_TRANSFORMER_H_INCLUDED
#define NNUE_FEATURE_TRANSFORMER_H_INCLUDED
#include "nnue_common.h"
#include "nnue_architecture.h"
#include <cstring> // std::memset()
#include <utility> // std::pair
namespace Stockfish::Eval::NNUE {
using BiasType = std::int16_t;
using WeightType = std::int16_t;
using PSQTWeightType = std::int32_t;
// If vector instructions are enabled, we update and refresh the
// accumulator tile by tile such that each tile fits in the CPU's
// vector registers.
#define VECTOR
static_assert(PSQTBuckets % 8 == 0,
"Per feature PSQT values cannot be processed at granularity lower than 8 at a time.");
#ifdef USE_AVX512
using vec_t = __m512i;
using psqt_vec_t = __m256i;
#define vec_load(a) _mm512_load_si512(a)
#define vec_store(a,b) _mm512_store_si512(a,b)
#define vec_add_16(a,b) _mm512_add_epi16(a,b)
#define vec_sub_16(a,b) _mm512_sub_epi16(a,b)
#define vec_mul_16(a,b) _mm512_mullo_epi16(a,b)
#define vec_zero() _mm512_setzero_epi32()
#define vec_set_16(a) _mm512_set1_epi16(a)
#define vec_max_16(a,b) _mm512_max_epi16(a,b)
#define vec_min_16(a,b) _mm512_min_epi16(a,b)
inline vec_t vec_msb_pack_16(vec_t a, vec_t b){
vec_t compacted = _mm512_packs_epi16(_mm512_srli_epi16(a,7),_mm512_srli_epi16(b,7));
return _mm512_permutexvar_epi64(_mm512_setr_epi64(0, 2, 4, 6, 1, 3, 5, 7), compacted);
}
#define vec_load_psqt(a) _mm256_load_si256(a)
#define vec_store_psqt(a,b) _mm256_store_si256(a,b)
#define vec_add_psqt_32(a,b) _mm256_add_epi32(a,b)
#define vec_sub_psqt_32(a,b) _mm256_sub_epi32(a,b)
#define vec_zero_psqt() _mm256_setzero_si256()
#define NumRegistersSIMD 32
#define MaxChunkSize 64
#elif USE_AVX2
using vec_t = __m256i;
using psqt_vec_t = __m256i;
#define vec_load(a) _mm256_load_si256(a)
#define vec_store(a,b) _mm256_store_si256(a,b)
#define vec_add_16(a,b) _mm256_add_epi16(a,b)
#define vec_sub_16(a,b) _mm256_sub_epi16(a,b)
#define vec_mul_16(a,b) _mm256_mullo_epi16(a,b)
#define vec_zero() _mm256_setzero_si256()
#define vec_set_16(a) _mm256_set1_epi16(a)
#define vec_max_16(a,b) _mm256_max_epi16(a,b)
#define vec_min_16(a,b) _mm256_min_epi16(a,b)
inline vec_t vec_msb_pack_16(vec_t a, vec_t b){
vec_t compacted = _mm256_packs_epi16(_mm256_srli_epi16(a,7), _mm256_srli_epi16(b,7));
return _mm256_permute4x64_epi64(compacted, 0b11011000);
}
#define vec_load_psqt(a) _mm256_load_si256(a)
#define vec_store_psqt(a,b) _mm256_store_si256(a,b)
#define vec_add_psqt_32(a,b) _mm256_add_epi32(a,b)
#define vec_sub_psqt_32(a,b) _mm256_sub_epi32(a,b)
#define vec_zero_psqt() _mm256_setzero_si256()
#define NumRegistersSIMD 16
#define MaxChunkSize 32
#elif USE_SSE2
using vec_t = __m128i;
using psqt_vec_t = __m128i;
#define vec_load(a) (*(a))
#define vec_store(a,b) *(a)=(b)
#define vec_add_16(a,b) _mm_add_epi16(a,b)
#define vec_sub_16(a,b) _mm_sub_epi16(a,b)
#define vec_mul_16(a,b) _mm_mullo_epi16(a,b)
#define vec_zero() _mm_setzero_si128()
#define vec_set_16(a) _mm_set1_epi16(a)
#define vec_max_16(a,b) _mm_max_epi16(a,b)
#define vec_min_16(a,b) _mm_min_epi16(a,b)
#define vec_msb_pack_16(a,b) _mm_packs_epi16(_mm_srli_epi16(a,7),_mm_srli_epi16(b,7))
#define vec_load_psqt(a) (*(a))
#define vec_store_psqt(a,b) *(a)=(b)
#define vec_add_psqt_32(a,b) _mm_add_epi32(a,b)
#define vec_sub_psqt_32(a,b) _mm_sub_epi32(a,b)
#define vec_zero_psqt() _mm_setzero_si128()
#define NumRegistersSIMD (Is64Bit ? 16 : 8)
#define MaxChunkSize 16
#elif USE_MMX
using vec_t = __m64;
using psqt_vec_t = __m64;
#define vec_load(a) (*(a))
#define vec_store(a,b) *(a)=(b)
#define vec_add_16(a,b) _mm_add_pi16(a,b)
#define vec_sub_16(a,b) _mm_sub_pi16(a,b)
#define vec_mul_16(a,b) _mm_mullo_pi16(a,b)
#define vec_zero() _mm_setzero_si64()
#define vec_set_16(a) _mm_set1_pi16(a)
inline vec_t vec_max_16(vec_t a,vec_t b){
vec_t comparison = _mm_cmpgt_pi16(a,b);
return _mm_or_si64(_mm_and_si64(comparison, a), _mm_andnot_si64(comparison, b));
}
inline vec_t vec_min_16(vec_t a,vec_t b){
vec_t comparison = _mm_cmpgt_pi16(a,b);
return _mm_or_si64(_mm_and_si64(comparison, b), _mm_andnot_si64(comparison, a));
}
#define vec_msb_pack_16(a,b) _mm_packs_pi16(_mm_srli_pi16(a,7),_mm_srli_pi16(b,7))
#define vec_load_psqt(a) (*(a))
#define vec_store_psqt(a,b) *(a)=(b)
#define vec_add_psqt_32(a,b) _mm_add_pi32(a,b)
#define vec_sub_psqt_32(a,b) _mm_sub_pi32(a,b)
#define vec_zero_psqt() _mm_setzero_si64()
#define vec_cleanup() _mm_empty()
#define NumRegistersSIMD 8
#define MaxChunkSize 8
#elif USE_NEON
using vec_t = int16x8_t;
using psqt_vec_t = int32x4_t;
#define vec_load(a) (*(a))
#define vec_store(a,b) *(a)=(b)
#define vec_add_16(a,b) vaddq_s16(a,b)
#define vec_sub_16(a,b) vsubq_s16(a,b)
#define vec_mul_16(a,b) vmulq_s16(a,b)
#define vec_zero() vec_t{0}
#define vec_set_16(a) vdupq_n_s16(a)
#define vec_max_16(a,b) vmaxq_s16(a,b)
#define vec_min_16(a,b) vminq_s16(a,b)
inline vec_t vec_msb_pack_16(vec_t a, vec_t b){
const int8x8_t shifta = vshrn_n_s16(a, 7);
const int8x8_t shiftb = vshrn_n_s16(b, 7);
const int8x16_t compacted = vcombine_s8(shifta,shiftb);
return *reinterpret_cast<const vec_t*> (&compacted);
}
#define vec_load_psqt(a) (*(a))
#define vec_store_psqt(a,b) *(a)=(b)
#define vec_add_psqt_32(a,b) vaddq_s32(a,b)
#define vec_sub_psqt_32(a,b) vsubq_s32(a,b)
#define vec_zero_psqt() psqt_vec_t{0}
#define NumRegistersSIMD 16
#define MaxChunkSize 16
#else
#undef VECTOR
#endif
#ifdef VECTOR
// Compute optimal SIMD register count for feature transformer accumulation.
// We use __m* types as template arguments, which causes GCC to emit warnings
// about losing some attribute information. This is irrelevant to us as we
// only take their size, so the following pragma are harmless.
#if defined(__GNUC__)
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wignored-attributes"
#endif
template <typename SIMDRegisterType,
typename LaneType,
int NumLanes,
int MaxRegisters>
static constexpr int BestRegisterCount()
{
#define RegisterSize sizeof(SIMDRegisterType)
#define LaneSize sizeof(LaneType)
static_assert(RegisterSize >= LaneSize);
static_assert(MaxRegisters <= NumRegistersSIMD);
static_assert(MaxRegisters > 0);
static_assert(NumRegistersSIMD > 0);
static_assert(RegisterSize % LaneSize == 0);
static_assert((NumLanes * LaneSize) % RegisterSize == 0);
const int ideal = (NumLanes * LaneSize) / RegisterSize;
if (ideal <= MaxRegisters)
return ideal;
// Look for the largest divisor of the ideal register count that is smaller than MaxRegisters
for (int divisor = MaxRegisters; divisor > 1; --divisor)
if (ideal % divisor == 0)
return divisor;
return 1;
}
static constexpr int NumRegs = BestRegisterCount<vec_t, WeightType, TransformedFeatureDimensions, NumRegistersSIMD>();
static constexpr int NumPsqtRegs = BestRegisterCount<psqt_vec_t, PSQTWeightType, PSQTBuckets, NumRegistersSIMD>();
#if defined(__GNUC__)
#pragma GCC diagnostic pop
#endif
#endif
// Input feature converter
class FeatureTransformer {
private:
// Number of output dimensions for one side
static constexpr IndexType HalfDimensions = TransformedFeatureDimensions;
#ifdef VECTOR
static constexpr IndexType TileHeight = NumRegs * sizeof(vec_t) / 2;
static constexpr IndexType PsqtTileHeight = NumPsqtRegs * sizeof(psqt_vec_t) / 4;
static_assert(HalfDimensions % TileHeight == 0, "TileHeight must divide HalfDimensions");
static_assert(PSQTBuckets % PsqtTileHeight == 0, "PsqtTileHeight must divide PSQTBuckets");
#endif
public:
// Output type
using OutputType = TransformedFeatureType;
// Number of input/output dimensions
static constexpr IndexType InputDimensions = FeatureSet::Dimensions;
static constexpr IndexType OutputDimensions = HalfDimensions;
// Size of forward propagation buffer
static constexpr std::size_t BufferSize =
OutputDimensions * sizeof(OutputType);
// Hash value embedded in the evaluation file
static constexpr std::uint32_t get_hash_value() {
return FeatureSet::HashValue ^ (OutputDimensions * 2);
}
// Read network parameters
bool read_parameters(std::istream& stream) {
read_leb_128<BiasType >(stream, biases , HalfDimensions );
read_leb_128<WeightType >(stream, weights , HalfDimensions * InputDimensions);
read_leb_128<PSQTWeightType>(stream, psqtWeights, PSQTBuckets * InputDimensions);
return !stream.fail();
}
// Write network parameters
bool write_parameters(std::ostream& stream) const {
write_leb_128<BiasType >(stream, biases , HalfDimensions );
write_leb_128<WeightType >(stream, weights , HalfDimensions * InputDimensions);
write_leb_128<PSQTWeightType>(stream, psqtWeights, PSQTBuckets * InputDimensions);
return !stream.fail();
}
// Convert input features
std::int32_t transform(const Position& pos, OutputType* output, int bucket) const {
update_accumulator<WHITE>(pos);
update_accumulator<BLACK>(pos);
const Color perspectives[2] = {pos.side_to_move(), ~pos.side_to_move()};
const auto& accumulation = pos.state()->accumulator.accumulation;
const auto& psqtAccumulation = pos.state()->accumulator.psqtAccumulation;
const auto psqt = (
psqtAccumulation[perspectives[0]][bucket]
- psqtAccumulation[perspectives[1]][bucket]
) / 2;
for (IndexType p = 0; p < 2; ++p)
{
const IndexType offset = (HalfDimensions / 2) * p;
#if defined(VECTOR)
constexpr IndexType OutputChunkSize = MaxChunkSize;
static_assert((HalfDimensions / 2) % OutputChunkSize == 0);
constexpr IndexType NumOutputChunks = HalfDimensions / 2 / OutputChunkSize;
vec_t Zero = vec_zero();
vec_t One = vec_set_16(127);
const vec_t* in0 = reinterpret_cast<const vec_t*>(&(accumulation[perspectives[p]][0]));
const vec_t* in1 = reinterpret_cast<const vec_t*>(&(accumulation[perspectives[p]][HalfDimensions / 2]));
vec_t* out = reinterpret_cast< vec_t*>(output + offset);
for (IndexType j = 0; j < NumOutputChunks; j += 1)
{
const vec_t sum0a = vec_max_16(vec_min_16(in0[j * 2 + 0], One), Zero);
const vec_t sum0b = vec_max_16(vec_min_16(in0[j * 2 + 1], One), Zero);
const vec_t sum1a = vec_max_16(vec_min_16(in1[j * 2 + 0], One), Zero);
const vec_t sum1b = vec_max_16(vec_min_16(in1[j * 2 + 1], One), Zero);
const vec_t pa = vec_mul_16(sum0a, sum1a);
const vec_t pb = vec_mul_16(sum0b, sum1b);
out[j] = vec_msb_pack_16(pa, pb);
}
#else
for (IndexType j = 0; j < HalfDimensions / 2; ++j) {
BiasType sum0 = accumulation[static_cast<int>(perspectives[p])][j + 0];
BiasType sum1 = accumulation[static_cast<int>(perspectives[p])][j + HalfDimensions / 2];
sum0 = std::max<int>(0, std::min<int>(127, sum0));
sum1 = std::max<int>(0, std::min<int>(127, sum1));
output[offset + j] = static_cast<OutputType>(sum0 * sum1 / 128);
}
#endif
}
#if defined(vec_cleanup)
vec_cleanup();
#endif
return psqt;
} // end of function transform()
void hint_common_access(const Position& pos) const {
hint_common_access_for_perspective<WHITE>(pos);
hint_common_access_for_perspective<BLACK>(pos);
}
private:
template<Color Perspective>
[[nodiscard]] std::pair<StateInfo*, StateInfo*> try_find_computed_accumulator(const Position& pos) const {
// Look for a usable accumulator of an earlier position. We keep track
// of the estimated gain in terms of features to be added/subtracted.
StateInfo *st = pos.state(), *next = nullptr;
int gain = FeatureSet::refresh_cost(pos);
while (st->previous && !st->accumulator.computed[Perspective])
{
// This governs when a full feature refresh is needed and how many
// updates are better than just one full refresh.
if ( FeatureSet::requires_refresh(st, Perspective)
|| (gain -= FeatureSet::update_cost(st) + 1) < 0)
break;
next = st;
st = st->previous;
}
return { st, next };
}
// NOTE: The parameter states_to_update is an array of position states, ending with nullptr.
// All states must be sequential, that is states_to_update[i] must either be reachable
// by repeatedly applying ->previous from states_to_update[i+1] or states_to_update[i] == nullptr.
// computed_st must be reachable by repeatedly applying ->previous on states_to_update[0], if not nullptr.
template<Color Perspective, size_t N>
void update_accumulator_incremental(const Position& pos, StateInfo* computed_st, StateInfo* states_to_update[N]) const {
static_assert(N > 0);
assert(states_to_update[N-1] == nullptr);
#ifdef VECTOR
// Gcc-10.2 unnecessarily spills AVX2 registers if this array
// is defined in the VECTOR code below, once in each branch
vec_t acc[NumRegs];
psqt_vec_t psqt[NumPsqtRegs];
#endif
if (states_to_update[0] == nullptr)
return;
// Update incrementally going back through states_to_update.
// Gather all features to be updated.
const Square ksq = pos.square<KING>(Perspective);
// The size must be enough to contain the largest possible update.
// That might depend on the feature set and generally relies on the
// feature set's update cost calculation to be correct and never
// allow updates with more added/removed features than MaxActiveDimensions.
FeatureSet::IndexList removed[N-1], added[N-1];
{
int i = N-2; // last potential state to update. Skip last element because it must be nullptr.
while (states_to_update[i] == nullptr)
--i;
StateInfo *st2 = states_to_update[i];
for (; i >= 0; --i)
{
states_to_update[i]->accumulator.computed[Perspective] = true;
StateInfo* end_state = i == 0 ? computed_st : states_to_update[i - 1];
for (; st2 != end_state; st2 = st2->previous)
FeatureSet::append_changed_indices<Perspective>(
ksq, st2->dirtyPiece, removed[i], added[i]);
}
}
StateInfo* st = computed_st;
// Now update the accumulators listed in states_to_update[], where the last element is a sentinel.
#ifdef VECTOR
for (IndexType j = 0; j < HalfDimensions / TileHeight; ++j)
{
// Load accumulator
auto accTile = reinterpret_cast<vec_t*>(
&st->accumulator.accumulation[Perspective][j * TileHeight]);
for (IndexType k = 0; k < NumRegs; ++k)
acc[k] = vec_load(&accTile[k]);
for (IndexType i = 0; states_to_update[i]; ++i)
{
// Difference calculation for the deactivated features
for (const auto index : removed[i])
{
const IndexType offset = HalfDimensions * index + j * TileHeight;
auto column = reinterpret_cast<const vec_t*>(&weights[offset]);
for (IndexType k = 0; k < NumRegs; ++k)
acc[k] = vec_sub_16(acc[k], column[k]);
}
// Difference calculation for the activated features
for (const auto index : added[i])
{
const IndexType offset = HalfDimensions * index + j * TileHeight;
auto column = reinterpret_cast<const vec_t*>(&weights[offset]);
for (IndexType k = 0; k < NumRegs; ++k)
acc[k] = vec_add_16(acc[k], column[k]);
}
// Store accumulator
accTile = reinterpret_cast<vec_t*>(
&states_to_update[i]->accumulator.accumulation[Perspective][j * TileHeight]);
for (IndexType k = 0; k < NumRegs; ++k)
vec_store(&accTile[k], acc[k]);
}
}
for (IndexType j = 0; j < PSQTBuckets / PsqtTileHeight; ++j)
{
// Load accumulator
auto accTilePsqt = reinterpret_cast<psqt_vec_t*>(
&st->accumulator.psqtAccumulation[Perspective][j * PsqtTileHeight]);
for (std::size_t k = 0; k < NumPsqtRegs; ++k)
psqt[k] = vec_load_psqt(&accTilePsqt[k]);
for (IndexType i = 0; states_to_update[i]; ++i)
{
// Difference calculation for the deactivated features
for (const auto index : removed[i])
{
const IndexType offset = PSQTBuckets * index + j * PsqtTileHeight;
auto columnPsqt = reinterpret_cast<const psqt_vec_t*>(&psqtWeights[offset]);
for (std::size_t k = 0; k < NumPsqtRegs; ++k)
psqt[k] = vec_sub_psqt_32(psqt[k], columnPsqt[k]);
}
// Difference calculation for the activated features
for (const auto index : added[i])
{
const IndexType offset = PSQTBuckets * index + j * PsqtTileHeight;
auto columnPsqt = reinterpret_cast<const psqt_vec_t*>(&psqtWeights[offset]);
for (std::size_t k = 0; k < NumPsqtRegs; ++k)
psqt[k] = vec_add_psqt_32(psqt[k], columnPsqt[k]);
}
// Store accumulator
accTilePsqt = reinterpret_cast<psqt_vec_t*>(
&states_to_update[i]->accumulator.psqtAccumulation[Perspective][j * PsqtTileHeight]);
for (std::size_t k = 0; k < NumPsqtRegs; ++k)
vec_store_psqt(&accTilePsqt[k], psqt[k]);
}
}
#else
for (IndexType i = 0; states_to_update[i]; ++i)
{
std::memcpy(states_to_update[i]->accumulator.accumulation[Perspective],
st->accumulator.accumulation[Perspective],
HalfDimensions * sizeof(BiasType));
for (std::size_t k = 0; k < PSQTBuckets; ++k)
states_to_update[i]->accumulator.psqtAccumulation[Perspective][k] = st->accumulator.psqtAccumulation[Perspective][k];
st = states_to_update[i];
// Difference calculation for the deactivated features
for (const auto index : removed[i])
{
const IndexType offset = HalfDimensions * index;
for (IndexType j = 0; j < HalfDimensions; ++j)
st->accumulator.accumulation[Perspective][j] -= weights[offset + j];
for (std::size_t k = 0; k < PSQTBuckets; ++k)
st->accumulator.psqtAccumulation[Perspective][k] -= psqtWeights[index * PSQTBuckets + k];
}
// Difference calculation for the activated features
for (const auto index : added[i])
{
const IndexType offset = HalfDimensions * index;
for (IndexType j = 0; j < HalfDimensions; ++j)
st->accumulator.accumulation[Perspective][j] += weights[offset + j];
for (std::size_t k = 0; k < PSQTBuckets; ++k)
st->accumulator.psqtAccumulation[Perspective][k] += psqtWeights[index * PSQTBuckets + k];
}
}
#endif
#if defined(USE_MMX)
_mm_empty();
#endif
}
template<Color Perspective>
void update_accumulator_refresh(const Position& pos) const {
#ifdef VECTOR
// Gcc-10.2 unnecessarily spills AVX2 registers if this array
// is defined in the VECTOR code below, once in each branch
vec_t acc[NumRegs];
psqt_vec_t psqt[NumPsqtRegs];
#endif
// Refresh the accumulator
// Could be extracted to a separate function because it's done in 2 places,
// but it's unclear if compilers would correctly handle register allocation.
auto& accumulator = pos.state()->accumulator;
accumulator.computed[Perspective] = true;
FeatureSet::IndexList active;
FeatureSet::append_active_indices<Perspective>(pos, active);
#ifdef VECTOR
for (IndexType j = 0; j < HalfDimensions / TileHeight; ++j)
{
auto biasesTile = reinterpret_cast<const vec_t*>(
&biases[j * TileHeight]);
for (IndexType k = 0; k < NumRegs; ++k)
acc[k] = biasesTile[k];
for (const auto index : active)
{
const IndexType offset = HalfDimensions * index + j * TileHeight;
auto column = reinterpret_cast<const vec_t*>(&weights[offset]);
for (unsigned k = 0; k < NumRegs; ++k)
acc[k] = vec_add_16(acc[k], column[k]);
}
auto accTile = reinterpret_cast<vec_t*>(
&accumulator.accumulation[Perspective][j * TileHeight]);
for (unsigned k = 0; k < NumRegs; k++)
vec_store(&accTile[k], acc[k]);
}
for (IndexType j = 0; j < PSQTBuckets / PsqtTileHeight; ++j)
{
for (std::size_t k = 0; k < NumPsqtRegs; ++k)
psqt[k] = vec_zero_psqt();
for (const auto index : active)
{
const IndexType offset = PSQTBuckets * index + j * PsqtTileHeight;
auto columnPsqt = reinterpret_cast<const psqt_vec_t*>(&psqtWeights[offset]);
for (std::size_t k = 0; k < NumPsqtRegs; ++k)
psqt[k] = vec_add_psqt_32(psqt[k], columnPsqt[k]);
}
auto accTilePsqt = reinterpret_cast<psqt_vec_t*>(
&accumulator.psqtAccumulation[Perspective][j * PsqtTileHeight]);
for (std::size_t k = 0; k < NumPsqtRegs; ++k)
vec_store_psqt(&accTilePsqt[k], psqt[k]);
}
#else
std::memcpy(accumulator.accumulation[Perspective], biases,
HalfDimensions * sizeof(BiasType));
for (std::size_t k = 0; k < PSQTBuckets; ++k)
accumulator.psqtAccumulation[Perspective][k] = 0;
for (const auto index : active)
{
const IndexType offset = HalfDimensions * index;
for (IndexType j = 0; j < HalfDimensions; ++j)
accumulator.accumulation[Perspective][j] += weights[offset + j];
for (std::size_t k = 0; k < PSQTBuckets; ++k)
accumulator.psqtAccumulation[Perspective][k] += psqtWeights[index * PSQTBuckets + k];
}
#endif
#if defined(USE_MMX)
_mm_empty();
#endif
}
template<Color Perspective>
void hint_common_access_for_perspective(const Position& pos) const {
// Works like update_accumulator, but performs less work.
// Updates ONLY the accumulator for pos.
// Look for a usable accumulator of an earlier position. We keep track
// of the estimated gain in terms of features to be added/subtracted.
// Fast early exit.
if (pos.state()->accumulator.computed[Perspective])
return;
auto [oldest_st, _] = try_find_computed_accumulator<Perspective>(pos);
if (oldest_st->accumulator.computed[Perspective])
{
// Only update current position accumulator to minimize work.
StateInfo* states_to_update[2] = { pos.state(), nullptr };
update_accumulator_incremental<Perspective, 2>(pos, oldest_st, states_to_update);
}
else
{
update_accumulator_refresh<Perspective>(pos);
}
}
template<Color Perspective>
void update_accumulator(const Position& pos) const {
auto [oldest_st, next] = try_find_computed_accumulator<Perspective>(pos);
if (oldest_st->accumulator.computed[Perspective])
{
if (next == nullptr)
return;
// Now update the accumulators listed in states_to_update[], where the last element is a sentinel.
// Currently we update 2 accumulators.
// 1. for the current position
// 2. the next accumulator after the computed one
// The heuristic may change in the future.
StateInfo *states_to_update[3] =
{ next, next == pos.state() ? nullptr : pos.state(), nullptr };
update_accumulator_incremental<Perspective, 3>(pos, oldest_st, states_to_update);
}
else
{
update_accumulator_refresh<Perspective>(pos);
}
}
alignas(CacheLineSize) BiasType biases[HalfDimensions];
alignas(CacheLineSize) WeightType weights[HalfDimensions * InputDimensions];
alignas(CacheLineSize) PSQTWeightType psqtWeights[InputDimensions * PSQTBuckets];
};
} // namespace Stockfish::Eval::NNUE
#endif // #ifndef NNUE_FEATURE_TRANSFORMER_H_INCLUDED

305
src/Stockfish/src/pawns.cpp Normal file
View file

@ -0,0 +1,305 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm>
#include <cassert>
#include "bitboard.h"
#include "pawns.h"
#include "position.h"
#include "thread.h"
namespace Stockfish {
namespace {
#define V Value
#define S(mg, eg) make_score(mg, eg)
// Pawn penalties
constexpr Score Backward = S( 6, 19);
constexpr Score Doubled = S(11, 51);
constexpr Score DoubledEarly = S(17, 7);
constexpr Score Isolated = S( 1, 20);
constexpr Score WeakLever = S( 2, 57);
constexpr Score WeakUnopposed = S(15, 18);
// Bonus for blocked pawns at 5th or 6th rank
constexpr Score BlockedPawn[2] = { S(-19, -8), S(-7, 3) };
constexpr Score BlockedStorm[RANK_NB] = {
S(0, 0), S(0, 0), S(64, 75), S(-3, 14), S(-12, 19), S(-7, 4), S(-10, 5)
};
// Connected pawn bonus
constexpr int Connected[RANK_NB] = { 0, 3, 7, 7, 15, 54, 86 };
// Strength of pawn shelter for our king by [distance from edge][rank].
// RANK_1 = 0 is used for files where we have no pawn, or pawn is behind our king.
constexpr Value ShelterStrength[int(FILE_NB) / 2][RANK_NB] = {
{ V(-2), V(85), V(95), V(53), V(39), V(23), V(25) },
{ V(-55), V(64), V(32), V(-55), V(-30), V(-11), V(-61) },
{ V(-11), V(75), V(19), V(-6), V(26), V(9), V(-47) },
{ V(-41), V(-11), V(-27), V(-58), V(-42), V(-66), V(-163) }
};
// Danger of enemy pawns moving toward our king by [distance from edge][rank].
// RANK_1 = 0 is used for files where the enemy has no pawn, or their pawn
// is behind our king. Note that UnblockedStorm[0][1-2] accommodate opponent pawn
// on edge, likely blocked by our king.
constexpr Value UnblockedStorm[int(FILE_NB) / 2][RANK_NB] = {
{ V(94), V(-280), V(-170), V(90), V(59), V(47), V(53) },
{ V(43), V(-17), V(128), V(39), V(26), V(-17), V(15) },
{ V(-9), V(62), V(170), V(34), V(-5), V(-20), V(-11) },
{ V(-27), V(-19), V(106), V(10), V(2), V(-13), V(-24) }
};
// KingOnFile[semi-open Us][semi-open Them] contains bonuses/penalties
// for king when the king is on a semi-open or open file.
constexpr Score KingOnFile[2][2] = {{ S(-18,11), S(-6,-3) },
{ S( 0, 0), S( 5,-4) }};
#undef S
#undef V
/// evaluate() calculates a score for the static pawn structure of the given position.
/// We cannot use the location of pieces or king in this function, as the evaluation
/// of the pawn structure will be stored in a small cache for speed reasons, and will
/// be re-used even when the pieces have moved.
template<Color Us>
Score evaluate(const Position& pos, Pawns::Entry* e) {
constexpr Color Them = ~Us;
constexpr Direction Up = pawn_push(Us);
constexpr Direction Down = -Up;
Bitboard neighbours, stoppers, support, phalanx, opposed;
Bitboard lever, leverPush, blocked;
Square s;
bool backward, passed, doubled;
Score score = SCORE_ZERO;
Bitboard b = pos.pieces(Us, PAWN);
Bitboard ourPawns = pos.pieces( Us, PAWN);
Bitboard theirPawns = pos.pieces(Them, PAWN);
Bitboard doubleAttackThem = pawn_double_attacks_bb<Them>(theirPawns);
e->passedPawns[Us] = 0;
e->kingSquares[Us] = SQ_NONE;
e->pawnAttacks[Us] = e->pawnAttacksSpan[Us] = pawn_attacks_bb<Us>(ourPawns);
e->blockedCount += popcount(shift<Up>(ourPawns) & (theirPawns | doubleAttackThem));
// Loop through all pawns of the current color and score each pawn
while (b)
{
s = pop_lsb(b);
assert(pos.piece_on(s) == make_piece(Us, PAWN));
Rank r = relative_rank(Us, s);
// Flag the pawn
opposed = theirPawns & forward_file_bb(Us, s);
blocked = theirPawns & (s + Up);
stoppers = theirPawns & passed_pawn_span(Us, s);
lever = theirPawns & pawn_attacks_bb(Us, s);
leverPush = theirPawns & pawn_attacks_bb(Us, s + Up);
doubled = ourPawns & (s - Up);
neighbours = ourPawns & adjacent_files_bb(s);
phalanx = neighbours & rank_bb(s);
support = neighbours & rank_bb(s - Up);
if (doubled)
{
// Additional doubled penalty if none of their pawns is fixed
if (!(ourPawns & shift<Down>(theirPawns | pawn_attacks_bb<Them>(theirPawns))))
score -= DoubledEarly;
}
// A pawn is backward when it is behind all pawns of the same color on
// the adjacent files and cannot safely advance.
backward = !(neighbours & forward_ranks_bb(Them, s + Up))
&& (leverPush | blocked);
// Compute additional span if pawn is not backward nor blocked
if (!backward && !blocked)
e->pawnAttacksSpan[Us] |= pawn_attack_span(Us, s);
// A pawn is passed if one of the three following conditions is true:
// (a) there is no stoppers except some levers
// (b) the only stoppers are the leverPush, but we outnumber them
// (c) there is only one front stopper which can be levered.
// (Refined in Evaluation::passed)
passed = !(stoppers ^ lever)
|| ( !(stoppers ^ leverPush)
&& popcount(phalanx) >= popcount(leverPush))
|| ( stoppers == blocked && r >= RANK_5
&& (shift<Up>(support) & ~(theirPawns | doubleAttackThem)));
passed &= !(forward_file_bb(Us, s) & ourPawns);
// Passed pawns will be properly scored later in evaluation when we have
// full attack info.
if (passed)
e->passedPawns[Us] |= s;
// Score this pawn
if (support | phalanx)
{
int v = Connected[r] * (2 + bool(phalanx) - bool(opposed))
+ 22 * popcount(support);
score += make_score(v, v * (r - 2) / 4);
}
else if (!neighbours)
{
if ( opposed
&& (ourPawns & forward_file_bb(Them, s))
&& !(theirPawns & adjacent_files_bb(s)))
score -= Doubled;
else
score -= Isolated
+ WeakUnopposed * !opposed;
}
else if (backward)
score -= Backward
+ WeakUnopposed * !opposed * bool(~(FileABB | FileHBB) & s);
if (!support)
score -= Doubled * doubled
+ WeakLever * more_than_one(lever);
if (blocked && r >= RANK_5)
score += BlockedPawn[r - RANK_5];
}
return score;
}
} // namespace
namespace Pawns {
/// Pawns::probe() looks up the current position's pawns configuration in
/// the pawns hash table. It returns a pointer to the Entry if the position
/// is found. Otherwise a new Entry is computed and stored there, so we don't
/// have to recompute all when the same pawns configuration occurs again.
Entry* probe(const Position& pos) {
Key key = pos.pawn_key();
Entry* e = pos.this_thread()->pawnsTable[key];
if (e->key == key)
return e;
e->key = key;
e->blockedCount = 0;
e->scores[WHITE] = evaluate<WHITE>(pos, e);
e->scores[BLACK] = evaluate<BLACK>(pos, e);
return e;
}
/// Entry::evaluate_shelter() calculates the shelter bonus and the storm
/// penalty for a king, looking at the king file and the two closest files.
template<Color Us>
Score Entry::evaluate_shelter(const Position& pos, Square ksq) const {
constexpr Color Them = ~Us;
Bitboard b = pos.pieces(PAWN) & ~forward_ranks_bb(Them, ksq);
Bitboard ourPawns = b & pos.pieces(Us) & ~pawnAttacks[Them];
Bitboard theirPawns = b & pos.pieces(Them);
Score bonus = make_score(5, 5);
File center = std::clamp(file_of(ksq), FILE_B, FILE_G);
for (File f = File(center - 1); f <= File(center + 1); ++f)
{
b = ourPawns & file_bb(f);
int ourRank = b ? relative_rank(Us, frontmost_sq(Them, b)) : 0;
b = theirPawns & file_bb(f);
int theirRank = b ? relative_rank(Us, frontmost_sq(Them, b)) : 0;
int d = edge_distance(f);
bonus += make_score(ShelterStrength[d][ourRank], 0);
if (ourRank && (ourRank == theirRank - 1))
bonus -= BlockedStorm[theirRank];
else
bonus -= make_score(UnblockedStorm[d][theirRank], 0);
}
// King On File
bonus -= KingOnFile[pos.is_on_semiopen_file(Us, ksq)][pos.is_on_semiopen_file(Them, ksq)];
return bonus;
}
/// Entry::do_king_safety() calculates a bonus for king safety. It is called only
/// when king square changes, which is about 20% of total king_safety() calls.
template<Color Us>
Score Entry::do_king_safety(const Position& pos) {
Square ksq = pos.square<KING>(Us);
kingSquares[Us] = ksq;
castlingRights[Us] = pos.castling_rights(Us);
auto compare = [](Score a, Score b) { return mg_value(a) < mg_value(b); };
Score shelter = evaluate_shelter<Us>(pos, ksq);
// If we can castle use the bonus after castling if it is bigger
if (pos.can_castle(Us & KING_SIDE))
shelter = std::max(shelter, evaluate_shelter<Us>(pos, relative_square(Us, SQ_G1)), compare);
if (pos.can_castle(Us & QUEEN_SIDE))
shelter = std::max(shelter, evaluate_shelter<Us>(pos, relative_square(Us, SQ_C1)), compare);
// In endgame we like to bring our king near our closest pawn
Bitboard pawns = pos.pieces(Us, PAWN);
int minPawnDist = 6;
if (pawns & attacks_bb<KING>(ksq))
minPawnDist = 1;
else while (pawns)
minPawnDist = std::min(minPawnDist, distance(ksq, pop_lsb(pawns)));
return shelter - make_score(0, 16 * minPawnDist);
}
// Explicit template instantiation
template Score Entry::do_king_safety<WHITE>(const Position& pos);
template Score Entry::do_king_safety<BLACK>(const Position& pos);
} // namespace Pawns
} // namespace Stockfish

70
src/Stockfish/src/pawns.h Normal file
View file

@ -0,0 +1,70 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef PAWNS_H_INCLUDED
#define PAWNS_H_INCLUDED
#include "misc.h"
#include "position.h"
#include "types.h"
namespace Stockfish::Pawns {
/// Pawns::Entry contains various information about a pawn structure. A lookup
/// to the pawn hash table (performed by calling the probe function) returns a
/// pointer to an Entry object.
struct Entry {
Score pawn_score(Color c) const { return scores[c]; }
Bitboard pawn_attacks(Color c) const { return pawnAttacks[c]; }
Bitboard passed_pawns(Color c) const { return passedPawns[c]; }
Bitboard pawn_attacks_span(Color c) const { return pawnAttacksSpan[c]; }
int passed_count() const { return popcount(passedPawns[WHITE] | passedPawns[BLACK]); }
int blocked_count() const { return blockedCount; }
template<Color Us>
Score king_safety(const Position& pos) {
return kingSquares[Us] == pos.square<KING>(Us) && castlingRights[Us] == pos.castling_rights(Us)
? kingSafety[Us] : (kingSafety[Us] = do_king_safety<Us>(pos));
}
template<Color Us>
Score do_king_safety(const Position& pos);
template<Color Us>
Score evaluate_shelter(const Position& pos, Square ksq) const;
Key key;
Score scores[COLOR_NB];
Bitboard passedPawns[COLOR_NB];
Bitboard pawnAttacks[COLOR_NB];
Bitboard pawnAttacksSpan[COLOR_NB];
Square kingSquares[COLOR_NB];
Score kingSafety[COLOR_NB];
int castlingRights[COLOR_NB];
int blockedCount;
};
using Table = HashTable<Entry, 131072>;
Entry* probe(const Position& pos);
} // namespace Stockfish::Pawns
#endif // #ifndef PAWNS_H_INCLUDED

1347
src/Stockfish/src/position.cpp Normal file
View file

File diff suppressed because it is too large Load diff

450
src/Stockfish/src/position.h Normal file
View file

@ -0,0 +1,450 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef POSITION_H_INCLUDED
#define POSITION_H_INCLUDED
#include <cassert>
#include <deque>
#include <memory> // For std::unique_ptr
#include <string>
#include "bitboard.h"
#include "evaluate.h"
#include "psqt.h"
#include "types.h"
#include "nnue/nnue_accumulator.h"
namespace Stockfish {
/// StateInfo struct stores information needed to restore a Position object to
/// its previous state when we retract a move. Whenever a move is made on the
/// board (by calling Position::do_move), a StateInfo object must be passed.
struct StateInfo {
// Copied when making a move
Key pawnKey;
Key materialKey;
Value nonPawnMaterial[COLOR_NB];
int castlingRights;
int rule50;
int pliesFromNull;
Square epSquare;
// Not copied when making a move (will be recomputed anyhow)
Key key;
Bitboard checkersBB;
StateInfo* previous;
Bitboard blockersForKing[COLOR_NB];
Bitboard pinners[COLOR_NB];
Bitboard checkSquares[PIECE_TYPE_NB];
Piece capturedPiece;
int repetition;
// Used by NNUE
Eval::NNUE::Accumulator accumulator;
DirtyPiece dirtyPiece;
};
/// A list to keep track of the position states along the setup moves (from the
/// start position to the position just before the search starts). Needed by
/// 'draw by repetition' detection. Use a std::deque because pointers to
/// elements are not invalidated upon list resizing.
using StateListPtr = std::unique_ptr<std::deque<StateInfo>>;
/// Position class stores information regarding the board representation as
/// pieces, side to move, hash keys, castling info, etc. Important methods are
/// do_move() and undo_move(), used by the search to update node info when
/// traversing the search tree.
class Thread;
class Position {
public:
static void init();
Position() = default;
Position(const Position&) = delete;
Position& operator=(const Position&) = delete;
// FEN string input/output
Position& set(const std::string& fenStr, bool isChess960, StateInfo* si, Thread* th);
Position& set(const std::string& code, Color c, StateInfo* si);
std::string fen() const;
// Position representation
Bitboard pieces(PieceType pt) const;
template<typename ...PieceTypes> Bitboard pieces(PieceType pt, PieceTypes... pts) const;
Bitboard pieces(Color c) const;
template<typename ...PieceTypes> Bitboard pieces(Color c, PieceTypes... pts) const;
Piece piece_on(Square s) const;
Square ep_square() const;
bool empty(Square s) const;
template<PieceType Pt> int count(Color c) const;
template<PieceType Pt> int count() const;
template<PieceType Pt> Square square(Color c) const;
bool is_on_semiopen_file(Color c, Square s) const;
// Castling
CastlingRights castling_rights(Color c) const;
bool can_castle(CastlingRights cr) const;
bool castling_impeded(CastlingRights cr) const;
Square castling_rook_square(CastlingRights cr) const;
// Checking
Bitboard checkers() const;
Bitboard blockers_for_king(Color c) const;
Bitboard check_squares(PieceType pt) const;
Bitboard pinners(Color c) const;
// Attacks to/from a given square
Bitboard attackers_to(Square s) const;
Bitboard attackers_to(Square s, Bitboard occupied) const;
Bitboard slider_blockers(Bitboard sliders, Square s, Bitboard& pinners) const;
template<PieceType Pt> Bitboard attacks_by(Color c) const;
// Properties of moves
bool legal(Move m) const;
bool pseudo_legal(const Move m) const;
bool capture(Move m) const;
bool capture_stage(Move m) const;
bool gives_check(Move m) const;
Piece moved_piece(Move m) const;
Piece captured_piece() const;
// Piece specific
bool pawn_passed(Color c, Square s) const;
bool opposite_bishops() const;
int pawns_on_same_color_squares(Color c, Square s) const;
// Doing and undoing moves
void do_move(Move m, StateInfo& newSt);
void do_move(Move m, StateInfo& newSt, bool givesCheck);
void undo_move(Move m);
void do_null_move(StateInfo& newSt);
void undo_null_move();
// Static Exchange Evaluation
bool see_ge(Move m, Value threshold = VALUE_ZERO) const;
bool see_ge(Move m, Bitboard& occupied, Value threshold = VALUE_ZERO) const;
// Accessing hash keys
Key key() const;
Key key_after(Move m) const;
Key material_key() const;
Key pawn_key() const;
// Other properties of the position
Color side_to_move() const;
int game_ply() const;
bool is_chess960() const;
Thread* this_thread() const;
bool is_draw(int ply) const;
bool has_game_cycle(int ply) const;
bool has_repeated() const;
int rule50_count() const;
Score psq_score() const;
Value psq_eg_stm() const;
Value non_pawn_material(Color c) const;
Value non_pawn_material() const;
// Position consistency check, for debugging
bool pos_is_ok() const;
void flip();
// Used by NNUE
StateInfo* state() const;
void put_piece(Piece pc, Square s);
void remove_piece(Square s);
private:
// Initialization helpers (used while setting up a position)
void set_castling_right(Color c, Square rfrom);
void set_state() const;
void set_check_info() const;
// Other helpers
void move_piece(Square from, Square to);
template<bool Do>
void do_castling(Color us, Square from, Square& to, Square& rfrom, Square& rto);
template<bool AfterMove>
Key adjust_key50(Key k) const;
// Data members
Piece board[SQUARE_NB];
Bitboard byTypeBB[PIECE_TYPE_NB];
Bitboard byColorBB[COLOR_NB];
int pieceCount[PIECE_NB];
int castlingRightsMask[SQUARE_NB];
Square castlingRookSquare[CASTLING_RIGHT_NB];
Bitboard castlingPath[CASTLING_RIGHT_NB];
Thread* thisThread;
StateInfo* st;
int gamePly;
Color sideToMove;
Score psq;
bool chess960;
};
std::ostream& operator<<(std::ostream& os, const Position& pos);
inline Color Position::side_to_move() const {
return sideToMove;
}
inline Piece Position::piece_on(Square s) const {
assert(is_ok(s));
return board[s];
}
inline bool Position::empty(Square s) const {
return piece_on(s) == NO_PIECE;
}
inline Piece Position::moved_piece(Move m) const {
return piece_on(from_sq(m));
}
inline Bitboard Position::pieces(PieceType pt = ALL_PIECES) const {
return byTypeBB[pt];
}
template<typename ...PieceTypes>
inline Bitboard Position::pieces(PieceType pt, PieceTypes... pts) const {
return pieces(pt) | pieces(pts...);
}
inline Bitboard Position::pieces(Color c) const {
return byColorBB[c];
}
template<typename ...PieceTypes>
inline Bitboard Position::pieces(Color c, PieceTypes... pts) const {
return pieces(c) & pieces(pts...);
}
template<PieceType Pt> inline int Position::count(Color c) const {
return pieceCount[make_piece(c, Pt)];
}
template<PieceType Pt> inline int Position::count() const {
return count<Pt>(WHITE) + count<Pt>(BLACK);
}
template<PieceType Pt> inline Square Position::square(Color c) const {
assert(count<Pt>(c) == 1);
return lsb(pieces(c, Pt));
}
inline Square Position::ep_square() const {
return st->epSquare;
}
inline bool Position::is_on_semiopen_file(Color c, Square s) const {
return !(pieces(c, PAWN) & file_bb(s));
}
inline bool Position::can_castle(CastlingRights cr) const {
return st->castlingRights & cr;
}
inline CastlingRights Position::castling_rights(Color c) const {
return c & CastlingRights(st->castlingRights);
}
inline bool Position::castling_impeded(CastlingRights cr) const {
assert(cr == WHITE_OO || cr == WHITE_OOO || cr == BLACK_OO || cr == BLACK_OOO);
return pieces() & castlingPath[cr];
}
inline Square Position::castling_rook_square(CastlingRights cr) const {
assert(cr == WHITE_OO || cr == WHITE_OOO || cr == BLACK_OO || cr == BLACK_OOO);
return castlingRookSquare[cr];
}
inline Bitboard Position::attackers_to(Square s) const {
return attackers_to(s, pieces());
}
template<PieceType Pt>
inline Bitboard Position::attacks_by(Color c) const {
if constexpr (Pt == PAWN)
return c == WHITE ? pawn_attacks_bb<WHITE>(pieces(WHITE, PAWN))
: pawn_attacks_bb<BLACK>(pieces(BLACK, PAWN));
else
{
Bitboard threats = 0;
Bitboard attackers = pieces(c, Pt);
while (attackers)
threats |= attacks_bb<Pt>(pop_lsb(attackers), pieces());
return threats;
}
}
inline Bitboard Position::checkers() const {
return st->checkersBB;
}
inline Bitboard Position::blockers_for_king(Color c) const {
return st->blockersForKing[c];
}
inline Bitboard Position::pinners(Color c) const {
return st->pinners[c];
}
inline Bitboard Position::check_squares(PieceType pt) const {
return st->checkSquares[pt];
}
inline bool Position::pawn_passed(Color c, Square s) const {
return !(pieces(~c, PAWN) & passed_pawn_span(c, s));
}
inline int Position::pawns_on_same_color_squares(Color c, Square s) const {
return popcount(pieces(c, PAWN) & ((DarkSquares & s) ? DarkSquares : ~DarkSquares));
}
inline Key Position::key() const {
return adjust_key50<false>(st->key);
}
template<bool AfterMove>
inline Key Position::adjust_key50(Key k) const
{
return st->rule50 < 14 - AfterMove
? k : k ^ make_key((st->rule50 - (14 - AfterMove)) / 8);
}
inline Key Position::pawn_key() const {
return st->pawnKey;
}
inline Key Position::material_key() const {
return st->materialKey;
}
inline Score Position::psq_score() const {
return psq;
}
inline Value Position::psq_eg_stm() const {
return (sideToMove == WHITE ? 1 : -1) * eg_value(psq);
}
inline Value Position::non_pawn_material(Color c) const {
return st->nonPawnMaterial[c];
}
inline Value Position::non_pawn_material() const {
return non_pawn_material(WHITE) + non_pawn_material(BLACK);
}
inline int Position::game_ply() const {
return gamePly;
}
inline int Position::rule50_count() const {
return st->rule50;
}
inline bool Position::opposite_bishops() const {
return count<BISHOP>(WHITE) == 1
&& count<BISHOP>(BLACK) == 1
&& opposite_colors(square<BISHOP>(WHITE), square<BISHOP>(BLACK));
}
inline bool Position::is_chess960() const {
return chess960;
}
inline bool Position::capture(Move m) const {
assert(is_ok(m));
return (!empty(to_sq(m)) && type_of(m) != CASTLING)
|| type_of(m) == EN_PASSANT;
}
// returns true if a move is generated from the capture stage
// having also queen promotions covered, i.e. consistency with the capture stage move generation
// is needed to avoid the generation of duplicate moves.
inline bool Position::capture_stage(Move m) const {
assert(is_ok(m));
return capture(m) || promotion_type(m) == QUEEN;
}
inline Piece Position::captured_piece() const {
return st->capturedPiece;
}
inline Thread* Position::this_thread() const {
return thisThread;
}
inline void Position::put_piece(Piece pc, Square s) {
board[s] = pc;
byTypeBB[ALL_PIECES] |= byTypeBB[type_of(pc)] |= s;
byColorBB[color_of(pc)] |= s;
pieceCount[pc]++;
pieceCount[make_piece(color_of(pc), ALL_PIECES)]++;
psq += PSQT::psq[pc][s];
}
inline void Position::remove_piece(Square s) {
Piece pc = board[s];
byTypeBB[ALL_PIECES] ^= s;
byTypeBB[type_of(pc)] ^= s;
byColorBB[color_of(pc)] ^= s;
board[s] = NO_PIECE;
pieceCount[pc]--;
pieceCount[make_piece(color_of(pc), ALL_PIECES)]--;
psq -= PSQT::psq[pc][s];
}
inline void Position::move_piece(Square from, Square to) {
Piece pc = board[from];
Bitboard fromTo = from | to;
byTypeBB[ALL_PIECES] ^= fromTo;
byTypeBB[type_of(pc)] ^= fromTo;
byColorBB[color_of(pc)] ^= fromTo;
board[from] = NO_PIECE;
board[to] = pc;
psq += PSQT::psq[pc][to] - PSQT::psq[pc][from];
}
inline void Position::do_move(Move m, StateInfo& newSt) {
do_move(m, newSt, gives_check(m));
}
inline StateInfo* Position::state() const {
return st;
}
} // namespace Stockfish
#endif // #ifndef POSITION_H_INCLUDED

131
src/Stockfish/src/psqt.cpp Normal file
View file

@ -0,0 +1,131 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "psqt.h"
#include <algorithm>
#include "bitboard.h"
#include "types.h"
namespace Stockfish {
namespace
{
auto constexpr S = make_score;
// 'Bonus' contains Piece-Square parameters.
// Scores are explicit for files A to D, implicitly mirrored for E to H.
constexpr Score Bonus[][RANK_NB][int(FILE_NB) / 2] = {
{ },
{ },
{ // Knight
{ S(-175, -96), S(-92,-65), S(-74,-49), S(-73,-21) },
{ S( -77, -67), S(-41,-54), S(-27,-18), S(-15, 8) },
{ S( -61, -40), S(-17,-27), S( 6, -8), S( 12, 29) },
{ S( -35, -35), S( 8, -2), S( 40, 13), S( 49, 28) },
{ S( -34, -45), S( 13,-16), S( 44, 9), S( 51, 39) },
{ S( -9, -51), S( 22,-44), S( 58,-16), S( 53, 17) },
{ S( -67, -69), S(-27,-50), S( 4,-51), S( 37, 12) },
{ S(-201,-100), S(-83,-88), S(-56,-56), S(-26,-17) }
},
{ // Bishop
{ S(-37,-40), S(-4 ,-21), S( -6,-26), S(-16, -8) },
{ S(-11,-26), S( 6, -9), S( 13,-12), S( 3, 1) },
{ S(-5 ,-11), S( 15, -1), S( -4, -1), S( 12, 7) },
{ S(-4 ,-14), S( 8, -4), S( 18, 0), S( 27, 12) },
{ S(-8 ,-12), S( 20, -1), S( 15,-10), S( 22, 11) },
{ S(-11,-21), S( 4, 4), S( 1, 3), S( 8, 4) },
{ S(-12,-22), S(-10,-14), S( 4, -1), S( 0, 1) },
{ S(-34,-32), S( 1,-29), S(-10,-26), S(-16,-17) }
},
{ // Rook
{ S(-31, -9), S(-20,-13), S(-14,-10), S(-5, -9) },
{ S(-21,-12), S(-13, -9), S( -8, -1), S( 6, -2) },
{ S(-25, 6), S(-11, -8), S( -1, -2), S( 3, -6) },
{ S(-13, -6), S( -5, 1), S( -4, -9), S(-6, 7) },
{ S(-27, -5), S(-15, 8), S( -4, 7), S( 3, -6) },
{ S(-22, 6), S( -2, 1), S( 6, -7), S(12, 10) },
{ S( -2, 4), S( 12, 5), S( 16, 20), S(18, -5) },
{ S(-17, 18), S(-19, 0), S( -1, 19), S( 9, 13) }
},
{ // Queen
{ S( 3,-69), S(-5,-57), S(-5,-47), S( 4,-26) },
{ S(-3,-54), S( 5,-31), S( 8,-22), S(12, -4) },
{ S(-3,-39), S( 6,-18), S(13, -9), S( 7, 3) },
{ S( 4,-23), S( 5, -3), S( 9, 13), S( 8, 24) },
{ S( 0,-29), S(14, -6), S(12, 9), S( 5, 21) },
{ S(-4,-38), S(10,-18), S( 6,-11), S( 8, 1) },
{ S(-5,-50), S( 6,-27), S(10,-24), S( 8, -8) },
{ S(-2,-74), S(-2,-52), S( 1,-43), S(-2,-34) }
},
{ // King
{ S(271, 1), S(327, 45), S(271, 85), S(198, 76) },
{ S(278, 53), S(303,100), S(234,133), S(179,135) },
{ S(195, 88), S(258,130), S(169,169), S(120,175) },
{ S(164,103), S(190,156), S(138,172), S( 98,172) },
{ S(154, 96), S(179,166), S(105,199), S( 70,199) },
{ S(123, 92), S(145,172), S( 81,184), S( 31,191) },
{ S( 88, 47), S(120,121), S( 65,116), S( 33,131) },
{ S( 59, 11), S( 89, 59), S( 45, 73), S( -1, 78) }
}
};
constexpr Score PBonus[RANK_NB][FILE_NB] =
{ // Pawn (asymmetric distribution)
{ },
{ S( 2, -8), S( 4, -6), S( 11, 9), S( 18, 5), S( 16, 16), S( 21, 6), S( 9, -6), S( -3,-18) },
{ S( -9, -9), S(-15, -7), S( 11,-10), S( 15, 5), S( 31, 2), S( 23, 3), S( 6, -8), S(-20, -5) },
{ S( -3, 7), S(-20, 1), S( 8, -8), S( 19, -2), S( 39,-14), S( 17,-13), S( 2,-11), S( -5, -6) },
{ S( 11, 12), S( -4, 6), S(-11, 2), S( 2, -6), S( 11, -5), S( 0, -4), S(-12, 14), S( 5, 9) },
{ S( 3, 27), S(-11, 18), S( -6, 19), S( 22, 29), S( -8, 30), S( -5, 9), S(-14, 8), S(-11, 14) },
{ S( -7, -1), S( 6,-14), S( -2, 13), S(-11, 22), S( 4, 24), S(-14, 17), S( 10, 7), S( -9, 7) }
};
} // namespace
namespace PSQT
{
Score psq[PIECE_NB][SQUARE_NB];
// PSQT::init() initializes piece-square tables: the white halves of the tables are
// copied from Bonus[] and PBonus[], adding the piece value, then the black halves of
// the tables are initialized by flipping and changing the sign of the white scores.
void init() {
for (Piece pc : {W_PAWN, W_KNIGHT, W_BISHOP, W_ROOK, W_QUEEN, W_KING})
{
Score score = make_score(PieceValue[MG][pc], PieceValue[EG][pc]);
for (Square s = SQ_A1; s <= SQ_H8; ++s)
{
File f = File(edge_distance(file_of(s)));
psq[ pc][s] = score + (type_of(pc) == PAWN ? PBonus[rank_of(s)][file_of(s)]
: Bonus[pc][rank_of(s)][f]);
psq[~pc][flip_rank(s)] = -psq[pc][s];
}
}
}
} // namespace PSQT
} // namespace Stockfish

38
src/Stockfish/src/psqt.h Normal file
View file

@ -0,0 +1,38 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef PSQT_H_INCLUDED
#define PSQT_H_INCLUDED
#include "types.h"
namespace Stockfish::PSQT
{
extern Score psq[PIECE_NB][SQUARE_NB];
// Fill psqt array from a set of internally linked parameters
void init();
} // namespace Stockfish::PSQT
#endif // PSQT_H_INCLUDED

1988
src/Stockfish/src/search.cpp Normal file
View file

File diff suppressed because it is too large Load diff

116
src/Stockfish/src/search.h Normal file
View file

@ -0,0 +1,116 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef SEARCH_H_INCLUDED
#define SEARCH_H_INCLUDED
#include <vector>
#include "misc.h"
#include "movepick.h"
#include "types.h"
namespace Stockfish {
class Position;
namespace Search {
/// Stack struct keeps track of the information we need to remember from nodes
/// shallower and deeper in the tree during the search. Each search thread has
/// its own array of Stack objects, indexed by the current ply.
struct Stack {
Move* pv;
PieceToHistory* continuationHistory;
int ply;
Move currentMove;
Move excludedMove;
Move killers[2];
Value staticEval;
int statScore;
int moveCount;
bool inCheck;
bool ttPv;
bool ttHit;
int doubleExtensions;
int cutoffCnt;
};
/// RootMove struct is used for moves at the root of the tree. For each root move
/// we store a score and a PV (really a refutation in the case of moves which
/// fail low). Score is normally set at -VALUE_INFINITE for all non-pv moves.
struct RootMove {
explicit RootMove(Move m) : pv(1, m) {}
bool extract_ponder_from_tt(Position& pos);
bool operator==(const Move& m) const { return pv[0] == m; }
bool operator<(const RootMove& m) const { // Sort in descending order
return m.score != score ? m.score < score
: m.previousScore < previousScore;
}
Value score = -VALUE_INFINITE;
Value previousScore = -VALUE_INFINITE;
Value averageScore = -VALUE_INFINITE;
Value uciScore = -VALUE_INFINITE;
bool scoreLowerbound = false;
bool scoreUpperbound = false;
int selDepth = 0;
int tbRank = 0;
Value tbScore;
std::vector<Move> pv;
};
using RootMoves = std::vector<RootMove>;
/// LimitsType struct stores information sent by GUI about available time to
/// search the current move, maximum depth/time, or if we are in analysis mode.
struct LimitsType {
LimitsType() { // Init explicitly due to broken value-initialization of non POD in MSVC
time[WHITE] = time[BLACK] = inc[WHITE] = inc[BLACK] = npmsec = movetime = TimePoint(0);
movestogo = depth = mate = perft = infinite = 0;
nodes = 0;
}
bool use_time_management() const {
return time[WHITE] || time[BLACK];
}
std::vector<Move> searchmoves;
TimePoint time[COLOR_NB], inc[COLOR_NB], npmsec, movetime, startTime;
int movestogo, depth, mate, perft, infinite;
int64_t nodes;
};
extern LimitsType Limits;
void init();
void clear();
} // namespace Search
} // namespace Stockfish
#endif // #ifndef SEARCH_H_INCLUDED

1626
src/Stockfish/src/syzygy/tbprobe.cpp Normal file
View file

File diff suppressed because it is too large Load diff

76
src/Stockfish/src/syzygy/tbprobe.h Normal file
View file

@ -0,0 +1,76 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TBPROBE_H
#define TBPROBE_H
#include <ostream>
#include "../search.h"
namespace Stockfish::Tablebases {
enum WDLScore {
WDLLoss = -2, // Loss
WDLBlessedLoss = -1, // Loss, but draw under 50-move rule
WDLDraw = 0, // Draw
WDLCursedWin = 1, // Win, but draw under 50-move rule
WDLWin = 2, // Win
};
// Possible states after a probing operation
enum ProbeState {
FAIL = 0, // Probe failed (missing file table)
OK = 1, // Probe successful
CHANGE_STM = -1, // DTZ should check the other side
ZEROING_BEST_MOVE = 2 // Best move zeroes DTZ (capture or pawn move)
};
extern int MaxCardinality;
void init(const std::string& paths);
WDLScore probe_wdl(Position& pos, ProbeState* result);
int probe_dtz(Position& pos, ProbeState* result);
bool root_probe(Position& pos, Search::RootMoves& rootMoves);
bool root_probe_wdl(Position& pos, Search::RootMoves& rootMoves);
void rank_root_moves(Position& pos, Search::RootMoves& rootMoves);
inline std::ostream& operator<<(std::ostream& os, const WDLScore v) {
os << (v == WDLLoss ? "Loss" :
v == WDLBlessedLoss ? "Blessed loss" :
v == WDLDraw ? "Draw" :
v == WDLCursedWin ? "Cursed win" :
v == WDLWin ? "Win" : "None");
return os;
}
inline std::ostream& operator<<(std::ostream& os, const ProbeState v) {
os << (v == FAIL ? "Failed" :
v == OK ? "Success" :
v == CHANGE_STM ? "Probed opponent side" :
v == ZEROING_BEST_MOVE ? "Best move zeroes DTZ" : "None");
return os;
}
} // namespace Stockfish::Tablebases
#endif

268
src/Stockfish/src/thread.cpp Normal file
View file

@ -0,0 +1,268 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cassert>
#include <algorithm> // For std::count
#include "movegen.h"
#include "search.h"
#include "thread.h"
#include "uci.h"
#include "syzygy/tbprobe.h"
#include "tt.h"
namespace Stockfish {
ThreadPool Threads; // Global object
/// Thread constructor launches the thread and waits until it goes to sleep
/// in idle_loop(). Note that 'searching' and 'exit' should be already set.
Thread::Thread(size_t n) : idx(n), stdThread(&Thread::idle_loop, this) {
wait_for_search_finished();
}
/// Thread destructor wakes up the thread in idle_loop() and waits
/// for its termination. Thread should be already waiting.
Thread::~Thread() {
assert(!searching);
exit = true;
start_searching();
stdThread.join();
}
/// Thread::clear() reset histories, usually before a new game
void Thread::clear() {
counterMoves.fill(MOVE_NONE);
mainHistory.fill(0);
captureHistory.fill(0);
for (bool inCheck : { false, true })
for (StatsType c : { NoCaptures, Captures })
for (auto& to : continuationHistory[inCheck][c])
for (auto& h : to)
h->fill(-71);
}
/// Thread::start_searching() wakes up the thread that will start the search
void Thread::start_searching() {
mutex.lock();
searching = true;
mutex.unlock(); // Unlock before notifying saves a few CPU-cycles
cv.notify_one(); // Wake up the thread in idle_loop()
}
/// Thread::wait_for_search_finished() blocks on the condition variable
/// until the thread has finished searching.
void Thread::wait_for_search_finished() {
std::unique_lock<std::mutex> lk(mutex);
cv.wait(lk, [&]{ return !searching; });
}
/// Thread::idle_loop() is where the thread is parked, blocked on the
/// condition variable, when it has no work to do.
void Thread::idle_loop() {
// If OS already scheduled us on a different group than 0 then don't overwrite
// the choice, eventually we are one of many one-threaded processes running on
// some Windows NUMA hardware, for instance in fishtest. To make it simple,
// just check if running threads are below a threshold, in this case all this
// NUMA machinery is not needed.
if (Options["Threads"] > 8)
WinProcGroup::bindThisThread(idx);
while (true)
{
std::unique_lock<std::mutex> lk(mutex);
searching = false;
cv.notify_one(); // Wake up anyone waiting for search finished
cv.wait(lk, [&]{ return searching; });
if (exit)
return;
lk.unlock();
search();
}
}
/// ThreadPool::set() creates/destroys threads to match the requested number.
/// Created and launched threads will immediately go to sleep in idle_loop.
/// Upon resizing, threads are recreated to allow for binding if necessary.
void ThreadPool::set(size_t requested) {
if (threads.size() > 0) // destroy any existing thread(s)
{
main()->wait_for_search_finished();
while (threads.size() > 0)
delete threads.back(), threads.pop_back();
}
if (requested > 0) // create new thread(s)
{
threads.push_back(new MainThread(0));
while (threads.size() < requested)
threads.push_back(new Thread(threads.size()));
clear();
// Reallocate the hash with the new threadpool size
TT.resize(size_t(Options["Hash"]));
// Init thread number dependent search params.
Search::init();
}
}
/// ThreadPool::clear() sets threadPool data to initial values
void ThreadPool::clear() {
for (Thread* th : threads)
th->clear();
main()->callsCnt = 0;
main()->bestPreviousScore = VALUE_INFINITE;
main()->bestPreviousAverageScore = VALUE_INFINITE;
main()->previousTimeReduction = 1.0;
}
/// ThreadPool::start_thinking() wakes up main thread waiting in idle_loop() and
/// returns immediately. Main thread will wake up other threads and start the search.
void ThreadPool::start_thinking(Position& pos, StateListPtr& states,
const Search::LimitsType& limits, bool ponderMode) {
main()->wait_for_search_finished();
main()->stopOnPonderhit = stop = false;
increaseDepth = true;
main()->ponder = ponderMode;
Search::Limits = limits;
Search::RootMoves rootMoves;
for (const auto& m : MoveList<LEGAL>(pos))
if ( limits.searchmoves.empty()
|| std::count(limits.searchmoves.begin(), limits.searchmoves.end(), m))
rootMoves.emplace_back(m);
if (!rootMoves.empty())
Tablebases::rank_root_moves(pos, rootMoves);
// After ownership transfer 'states' becomes empty, so if we stop the search
// and call 'go' again without setting a new position states.get() == nullptr.
assert(states.get() || setupStates.get());
if (states.get())
setupStates = std::move(states); // Ownership transfer, states is now empty
// We use Position::set() to set root position across threads. But there are
// some StateInfo fields (previous, pliesFromNull, capturedPiece) that cannot
// be deduced from a fen string, so set() clears them and they are set from
// setupStates->back() later. The rootState is per thread, earlier states are shared
// since they are read-only.
for (Thread* th : threads)
{
th->nodes = th->tbHits = th->nmpMinPly = th->bestMoveChanges = 0;
th->rootDepth = th->completedDepth = 0;
th->rootMoves = rootMoves;
th->rootPos.set(pos.fen(), pos.is_chess960(), &th->rootState, th);
th->rootState = setupStates->back();
}
main()->start_searching();
}
Thread* ThreadPool::get_best_thread() const {
Thread* bestThread = threads.front();
std::map<Move, int64_t> votes;
Value minScore = VALUE_NONE;
// Find minimum score of all threads
for (Thread* th: threads)
minScore = std::min(minScore, th->rootMoves[0].score);
// Vote according to score and depth, and select the best thread
auto thread_value = [minScore](Thread* th) {
return (th->rootMoves[0].score - minScore + 14) * int(th->completedDepth);
};
for (Thread* th : threads)
votes[th->rootMoves[0].pv[0]] += thread_value(th);
for (Thread* th : threads)
if (abs(bestThread->rootMoves[0].score) >= VALUE_TB_WIN_IN_MAX_PLY)
{
// Make sure we pick the shortest mate / TB conversion or stave off mate the longest
if (th->rootMoves[0].score > bestThread->rootMoves[0].score)
bestThread = th;
}
else if ( th->rootMoves[0].score >= VALUE_TB_WIN_IN_MAX_PLY
|| ( th->rootMoves[0].score > VALUE_TB_LOSS_IN_MAX_PLY
&& ( votes[th->rootMoves[0].pv[0]] > votes[bestThread->rootMoves[0].pv[0]]
|| ( votes[th->rootMoves[0].pv[0]] == votes[bestThread->rootMoves[0].pv[0]]
&& thread_value(th) * int(th->rootMoves[0].pv.size() > 2)
> thread_value(bestThread) * int(bestThread->rootMoves[0].pv.size() > 2)))))
bestThread = th;
return bestThread;
}
/// Start non-main threads
void ThreadPool::start_searching() {
for (Thread* th : threads)
if (th != threads.front())
th->start_searching();
}
/// Wait for non-main threads
void ThreadPool::wait_for_search_finished() const {
for (Thread* th : threads)
if (th != threads.front())
th->wait_for_search_finished();
}
} // namespace Stockfish

141
src/Stockfish/src/thread.h Normal file
View file

@ -0,0 +1,141 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef THREAD_H_INCLUDED
#define THREAD_H_INCLUDED
#include <atomic>
#include <condition_variable>
#include <mutex>
#include <thread>
#include <vector>
#include "material.h"
#include "movepick.h"
#include "pawns.h"
#include "position.h"
#include "search.h"
#include "thread_win32_osx.h"
namespace Stockfish {
/// Thread class keeps together all the thread-related stuff. We use
/// per-thread pawn and material hash tables so that once we get a
/// pointer to an entry its life time is unlimited and we don't have
/// to care about someone changing the entry under our feet.
class Thread {
std::mutex mutex;
std::condition_variable cv;
size_t idx;
bool exit = false, searching = true; // Set before starting std::thread
NativeThread stdThread;
public:
explicit Thread(size_t);
virtual ~Thread();
virtual void search();
void clear();
void idle_loop();
void start_searching();
void wait_for_search_finished();
size_t id() const { return idx; }
Pawns::Table pawnsTable;
Material::Table materialTable;
size_t pvIdx, pvLast;
std::atomic<uint64_t> nodes, tbHits, bestMoveChanges;
int selDepth, nmpMinPly;
Value bestValue, optimism[COLOR_NB];
Position rootPos;
StateInfo rootState;
Search::RootMoves rootMoves;
Depth rootDepth, completedDepth;
Value rootDelta;
CounterMoveHistory counterMoves;
ButterflyHistory mainHistory;
CapturePieceToHistory captureHistory;
ContinuationHistory continuationHistory[2][2];
};
/// MainThread is a derived class specific for main thread
struct MainThread : public Thread {
using Thread::Thread;
void search() override;
void check_time();
double previousTimeReduction;
Value bestPreviousScore;
Value bestPreviousAverageScore;
Value iterValue[4];
int callsCnt;
bool stopOnPonderhit;
std::atomic_bool ponder;
};
/// ThreadPool struct handles all the threads-related stuff like init, starting,
/// parking and, most importantly, launching a thread. All the access to threads
/// is done through this class.
struct ThreadPool {
void start_thinking(Position&, StateListPtr&, const Search::LimitsType&, bool = false);
void clear();
void set(size_t);
MainThread* main() const { return static_cast<MainThread*>(threads.front()); }
uint64_t nodes_searched() const { return accumulate(&Thread::nodes); }
uint64_t tb_hits() const { return accumulate(&Thread::tbHits); }
Thread* get_best_thread() const;
void start_searching();
void wait_for_search_finished() const;
std::atomic_bool stop, increaseDepth;
auto cbegin() const noexcept { return threads.cbegin(); }
auto begin() noexcept { return threads.begin(); }
auto end() noexcept { return threads.end(); }
auto cend() const noexcept { return threads.cend(); }
auto size() const noexcept { return threads.size(); }
auto empty() const noexcept { return threads.empty(); }
private:
StateListPtr setupStates;
std::vector<Thread*> threads;
uint64_t accumulate(std::atomic<uint64_t> Thread::* member) const {
uint64_t sum = 0;
for (Thread* th : threads)
sum += (th->*member).load(std::memory_order_relaxed);
return sum;
}
};
extern ThreadPool Threads;
} // namespace Stockfish
#endif // #ifndef THREAD_H_INCLUDED

74
src/Stockfish/src/thread_win32_osx.h Normal file
View file

@ -0,0 +1,74 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef THREAD_WIN32_OSX_H_INCLUDED
#define THREAD_WIN32_OSX_H_INCLUDED
#include <thread>
/// On OSX threads other than the main thread are created with a reduced stack
/// size of 512KB by default, this is too low for deep searches, which require
/// somewhat more than 1MB stack, so adjust it to TH_STACK_SIZE.
/// The implementation calls pthread_create() with the stack size parameter
/// equal to the linux 8MB default, on platforms that support it.
#if defined(__APPLE__) || defined(__MINGW32__) || defined(__MINGW64__) || defined(USE_PTHREADS)
#include <pthread.h>
namespace Stockfish {
static const size_t TH_STACK_SIZE = 8 * 1024 * 1024;
template <class T, class P = std::pair<T*, void(T::*)()>>
void* start_routine(void* ptr)
{
P* p = reinterpret_cast<P*>(ptr);
(p->first->*(p->second))(); // Call member function pointer
delete p;
return nullptr;
}
class NativeThread {
pthread_t thread;
public:
template<class T, class P = std::pair<T*, void(T::*)()>>
explicit NativeThread(void(T::*fun)(), T* obj) {
pthread_attr_t attr_storage, *attr = &attr_storage;
pthread_attr_init(attr);
pthread_attr_setstacksize(attr, TH_STACK_SIZE);
pthread_create(&thread, attr, start_routine<T>, new P(obj, fun));
}
void join() { pthread_join(thread, nullptr); }
};
} // namespace Stockfish
#else // Default case: use STL classes
namespace Stockfish {
using NativeThread = std::thread;
} // namespace Stockfish
#endif
#endif // #ifndef THREAD_WIN32_OSX_H_INCLUDED

109
src/Stockfish/src/timeman.cpp Normal file
View file

@ -0,0 +1,109 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm>
#include <cfloat>
#include <cmath>
#include "search.h"
#include "timeman.h"
#include "uci.h"
namespace Stockfish {
TimeManagement Time; // Our global time management object
/// TimeManagement::init() is called at the beginning of the search and calculates
/// the bounds of time allowed for the current game ply. We currently support:
// 1) x basetime (+ z increment)
// 2) x moves in y seconds (+ z increment)
void TimeManagement::init(Search::LimitsType& limits, Color us, int ply) {
// if we have no time, no need to initialize TM, except for the start time,
// which is used by movetime.
startTime = limits.startTime;
if (limits.time[us] == 0)
return;
TimePoint moveOverhead = TimePoint(Options["Move Overhead"]);
TimePoint slowMover = TimePoint(Options["Slow Mover"]);
TimePoint npmsec = TimePoint(Options["nodestime"]);
// optScale is a percentage of available time to use for the current move.
// maxScale is a multiplier applied to optimumTime.
double optScale, maxScale;
// If we have to play in 'nodes as time' mode, then convert from time
// to nodes, and use resulting values in time management formulas.
// WARNING: to avoid time losses, the given npmsec (nodes per millisecond)
// must be much lower than the real engine speed.
if (npmsec)
{
if (!availableNodes) // Only once at game start
availableNodes = npmsec * limits.time[us]; // Time is in msec
// Convert from milliseconds to nodes
limits.time[us] = TimePoint(availableNodes);
limits.inc[us] *= npmsec;
limits.npmsec = npmsec;
}
// Maximum move horizon of 50 moves
int mtg = limits.movestogo ? std::min(limits.movestogo, 50) : 50;
// Make sure timeLeft is > 0 since we may use it as a divisor
TimePoint timeLeft = std::max(TimePoint(1),
limits.time[us] + limits.inc[us] * (mtg - 1) - moveOverhead * (2 + mtg));
// Use extra time with larger increments
double optExtra = std::clamp(1.0 + 12.0 * limits.inc[us] / limits.time[us], 1.0, 1.12);
// A user may scale time usage by setting UCI option "Slow Mover"
// Default is 100 and changing this value will probably lose elo.
timeLeft = slowMover * timeLeft / 100;
// x basetime (+ z increment)
// If there is a healthy increment, timeLeft can exceed actual available
// game time for the current move, so also cap to 20% of available game time.
if (limits.movestogo == 0)
{
optScale = std::min(0.0120 + std::pow(ply + 3.0, 0.45) * 0.0039,
0.2 * limits.time[us] / double(timeLeft))
* optExtra;
maxScale = std::min(7.0, 4.0 + ply / 12.0);
}
// x moves in y seconds (+ z increment)
else
{
optScale = std::min((0.88 + ply / 116.4) / mtg,
0.88 * limits.time[us] / double(timeLeft));
maxScale = std::min(6.3, 1.5 + 0.11 * mtg);
}
// Never use more than 80% of the available time for this move
optimumTime = TimePoint(optScale * timeLeft);
maximumTime = TimePoint(std::min(0.8 * limits.time[us] - moveOverhead, maxScale * optimumTime));
if (Options["Ponder"])
optimumTime += optimumTime / 4;
}
} // namespace Stockfish

51
src/Stockfish/src/timeman.h Normal file
View file

@ -0,0 +1,51 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TIMEMAN_H_INCLUDED
#define TIMEMAN_H_INCLUDED
#include "misc.h"
#include "search.h"
#include "thread.h"
namespace Stockfish {
/// The TimeManagement class computes the optimal time to think depending on
/// the maximum available time, the game move number and other parameters.
class TimeManagement {
public:
void init(Search::LimitsType& limits, Color us, int ply);
TimePoint optimum() const { return optimumTime; }
TimePoint maximum() const { return maximumTime; }
TimePoint elapsed() const { return Search::Limits.npmsec ?
TimePoint(Threads.nodes_searched()) : now() - startTime; }
int64_t availableNodes; // When in 'nodes as time' mode
private:
TimePoint startTime;
TimePoint optimumTime;
TimePoint maximumTime;
};
extern TimeManagement Time;
} // namespace Stockfish
#endif // #ifndef TIMEMAN_H_INCLUDED

162
src/Stockfish/src/tt.cpp Normal file
View file

@ -0,0 +1,162 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cstring> // For std::memset
#include <iostream>
#include <thread>
#include "bitboard.h"
#include "misc.h"
#include "thread.h"
#include "tt.h"
#include "uci.h"
namespace Stockfish {
TranspositionTable TT; // Our global transposition table
/// TTEntry::save() populates the TTEntry with a new node's data, possibly
/// overwriting an old position. Update is not atomic and can be racy.
void TTEntry::save(Key k, Value v, bool pv, Bound b, Depth d, Move m, Value ev) {
// Preserve any existing move for the same position
if (m || (uint16_t)k != key16)
move16 = (uint16_t)m;
// Overwrite less valuable entries (cheapest checks first)
if ( b == BOUND_EXACT
|| (uint16_t)k != key16
|| d - DEPTH_OFFSET + 2 * pv > depth8 - 4)
{
assert(d > DEPTH_OFFSET);
assert(d < 256 + DEPTH_OFFSET);
key16 = (uint16_t)k;
depth8 = (uint8_t)(d - DEPTH_OFFSET);
genBound8 = (uint8_t)(TT.generation8 | uint8_t(pv) << 2 | b);
value16 = (int16_t)v;
eval16 = (int16_t)ev;
}
}
/// TranspositionTable::resize() sets the size of the transposition table,
/// measured in megabytes. Transposition table consists of a power of 2 number
/// of clusters and each cluster consists of ClusterSize number of TTEntry.
void TranspositionTable::resize(size_t mbSize) {
Threads.main()->wait_for_search_finished();
aligned_large_pages_free(table);
clusterCount = mbSize * 1024 * 1024 / sizeof(Cluster);
table = static_cast<Cluster*>(aligned_large_pages_alloc(clusterCount * sizeof(Cluster)));
if (!table)
{
std::cerr << "Failed to allocate " << mbSize
<< "MB for transposition table." << fakeendl;
exit(EXIT_FAILURE);
}
clear();
}
/// TranspositionTable::clear() initializes the entire transposition table to zero,
// in a multi-threaded way.
void TranspositionTable::clear() {
std::vector<std::thread> threads;
for (size_t idx = 0; idx < size_t(Options["Threads"]); ++idx)
{
threads.emplace_back([this, idx]() {
// Thread binding gives faster search on systems with a first-touch policy
if (Options["Threads"] > 8)
WinProcGroup::bindThisThread(idx);
// Each thread will zero its part of the hash table
const size_t stride = size_t(clusterCount / Options["Threads"]),
start = size_t(stride * idx),
len = idx != size_t(Options["Threads"]) - 1 ?
stride : clusterCount - start;
std::memset(&table[start], 0, len * sizeof(Cluster));
});
}
for (std::thread& th : threads)
th.join();
}
/// TranspositionTable::probe() looks up the current position in the transposition
/// table. It returns true and a pointer to the TTEntry if the position is found.
/// Otherwise, it returns false and a pointer to an empty or least valuable TTEntry
/// to be replaced later. The replace value of an entry is calculated as its depth
/// minus 8 times its relative age. TTEntry t1 is considered more valuable than
/// TTEntry t2 if its replace value is greater than that of t2.
TTEntry* TranspositionTable::probe(const Key key, bool& found) const {
TTEntry* const tte = first_entry(key);
const uint16_t key16 = (uint16_t)key; // Use the low 16 bits as key inside the cluster
for (int i = 0; i < ClusterSize; ++i)
if (tte[i].key16 == key16 || !tte[i].depth8)
{
tte[i].genBound8 = uint8_t(generation8 | (tte[i].genBound8 & (GENERATION_DELTA - 1))); // Refresh
return found = (bool)tte[i].depth8, &tte[i];
}
// Find an entry to be replaced according to the replacement strategy
TTEntry* replace = tte;
for (int i = 1; i < ClusterSize; ++i)
// Due to our packed storage format for generation and its cyclic
// nature we add GENERATION_CYCLE (256 is the modulus, plus what
// is needed to keep the unrelated lowest n bits from affecting
// the result) to calculate the entry age correctly even after
// generation8 overflows into the next cycle.
if ( replace->depth8 - ((GENERATION_CYCLE + generation8 - replace->genBound8) & GENERATION_MASK)
> tte[i].depth8 - ((GENERATION_CYCLE + generation8 - tte[i].genBound8) & GENERATION_MASK))
replace = &tte[i];
return found = false, replace;
}
/// TranspositionTable::hashfull() returns an approximation of the hashtable
/// occupation during a search. The hash is x permill full, as per UCI protocol.
int TranspositionTable::hashfull() const {
int cnt = 0;
for (int i = 0; i < 1000; ++i)
for (int j = 0; j < ClusterSize; ++j)
cnt += table[i].entry[j].depth8 && (table[i].entry[j].genBound8 & GENERATION_MASK) == generation8;
return cnt / ClusterSize;
}
} // namespace Stockfish

107
src/Stockfish/src/tt.h Normal file
View file

@ -0,0 +1,107 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TT_H_INCLUDED
#define TT_H_INCLUDED
#include "misc.h"
#include "types.h"
namespace Stockfish {
/// TTEntry struct is the 10 bytes transposition table entry, defined as below:
///
/// key 16 bit
/// depth 8 bit
/// generation 5 bit
/// pv node 1 bit
/// bound type 2 bit
/// move 16 bit
/// value 16 bit
/// eval value 16 bit
struct TTEntry {
Move move() const { return (Move )move16; }
Value value() const { return (Value)value16; }
Value eval() const { return (Value)eval16; }
Depth depth() const { return (Depth)depth8 + DEPTH_OFFSET; }
bool is_pv() const { return (bool)(genBound8 & 0x4); }
Bound bound() const { return (Bound)(genBound8 & 0x3); }
void save(Key k, Value v, bool pv, Bound b, Depth d, Move m, Value ev);
private:
friend class TranspositionTable;
uint16_t key16;
uint8_t depth8;
uint8_t genBound8;
uint16_t move16;
int16_t value16;
int16_t eval16;
};
/// A TranspositionTable is an array of Cluster, of size clusterCount. Each
/// cluster consists of ClusterSize number of TTEntry. Each non-empty TTEntry
/// contains information on exactly one position. The size of a Cluster should
/// divide the size of a cache line for best performance, as the cacheline is
/// prefetched when possible.
class TranspositionTable {
static constexpr int ClusterSize = 3;
struct Cluster {
TTEntry entry[ClusterSize];
char padding[2]; // Pad to 32 bytes
};
static_assert(sizeof(Cluster) == 32, "Unexpected Cluster size");
// Constants used to refresh the hash table periodically
static constexpr unsigned GENERATION_BITS = 3; // nb of bits reserved for other things
static constexpr int GENERATION_DELTA = (1 << GENERATION_BITS); // increment for generation field
static constexpr int GENERATION_CYCLE = 255 + (1 << GENERATION_BITS); // cycle length
static constexpr int GENERATION_MASK = (0xFF << GENERATION_BITS) & 0xFF; // mask to pull out generation number
public:
~TranspositionTable() { aligned_large_pages_free(table); }
void new_search() { generation8 += GENERATION_DELTA; } // Lower bits are used for other things
TTEntry* probe(const Key key, bool& found) const;
int hashfull() const;
void resize(size_t mbSize);
void clear();
TTEntry* first_entry(const Key key) const {
return &table[mul_hi64(key, clusterCount)].entry[0];
}
private:
friend struct TTEntry;
size_t clusterCount;
Cluster* table;
uint8_t generation8; // Size must be not bigger than TTEntry::genBound8
};
extern TranspositionTable TT;
} // namespace Stockfish
#endif // #ifndef TT_H_INCLUDED

133
src/Stockfish/src/tune.cpp Normal file
View file

@ -0,0 +1,133 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm>
#include <iostream>
#include <sstream>
#include "types.h"
#include "misc.h"
#include "uci.h"
using std::string;
namespace Stockfish {
bool Tune::update_on_last;
const UCI::Option* LastOption = nullptr;
static std::map<std::string, int> TuneResults;
string Tune::next(string& names, bool pop) {
string name;
do {
string token = names.substr(0, names.find(','));
if (pop)
names.erase(0, token.size() + 1);
std::stringstream ws(token);
name += (ws >> token, token); // Remove trailing whitespace
} while ( std::count(name.begin(), name.end(), '(')
- std::count(name.begin(), name.end(), ')'));
return name;
}
static void on_tune(const UCI::Option& o) {
if (!Tune::update_on_last || LastOption == &o)
Tune::read_options();
}
static void make_option(const string& n, int v, const SetRange& r) {
// Do not generate option when there is nothing to tune (ie. min = max)
if (r(v).first == r(v).second)
return;
if (TuneResults.count(n))
v = TuneResults[n];
Options[n] << UCI::Option(v, r(v).first, r(v).second, on_tune);
LastOption = &Options[n];
// Print formatted parameters, ready to be copy-pasted in Fishtest
fakeout << n << ","
<< v << ","
<< r(v).first << "," << r(v).second << ","
<< (r(v).second - r(v).first) / 20.0 << ","
<< "0.0020"
<< fakeendl;
}
template<> void Tune::Entry<int>::init_option() { make_option(name, value, range); }
template<> void Tune::Entry<int>::read_option() {
if (Options.count(name))
value = int(Options[name]);
}
template<> void Tune::Entry<Value>::init_option() { make_option(name, value, range); }
template<> void Tune::Entry<Value>::read_option() {
if (Options.count(name))
value = Value(int(Options[name]));
}
template<> void Tune::Entry<Score>::init_option() {
make_option("m" + name, mg_value(value), range);
make_option("e" + name, eg_value(value), range);
}
template<> void Tune::Entry<Score>::read_option() {
if (Options.count("m" + name))
value = make_score(int(Options["m" + name]), eg_value(value));
if (Options.count("e" + name))
value = make_score(mg_value(value), int(Options["e" + name]));
}
// Instead of a variable here we have a PostUpdate function: just call it
template<> void Tune::Entry<Tune::PostUpdate>::init_option() {}
template<> void Tune::Entry<Tune::PostUpdate>::read_option() { value(); }
} // namespace Stockfish
// Init options with tuning session results instead of default values. Useful to
// get correct bench signature after a tuning session or to test tuned values.
// Just copy fishtest tuning results in a result.txt file and extract the
// values with:
//
// cat results.txt | sed 's/^param: \([^,]*\), best: \([^,]*\).*/ TuneResults["\1"] = int(round(\2));/'
//
// Then paste the output below, as the function body
#include <cmath>
namespace Stockfish {
void Tune::read_results() {
/* ...insert your values here... */
}
} // namespace Stockfish

163
src/Stockfish/src/tune.h Normal file
View file

@ -0,0 +1,163 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TUNE_H_INCLUDED
#define TUNE_H_INCLUDED
#include <memory>
#include <string>
#include <type_traits>
#include <vector>
namespace Stockfish {
using Range = std::pair<int, int>; // Option's min-max values
using RangeFun = Range (int);
// Default Range function, to calculate Option's min-max values
inline Range default_range(int v) {
return v > 0 ? Range(0, 2 * v) : Range(2 * v, 0);
}
struct SetRange {
explicit SetRange(RangeFun f) : fun(f) {}
SetRange(int min, int max) : fun(nullptr), range(min, max) {}
Range operator()(int v) const { return fun ? fun(v) : range; }
RangeFun* fun;
Range range;
};
#define SetDefaultRange SetRange(default_range)
/// Tune class implements the 'magic' code that makes the setup of a fishtest
/// tuning session as easy as it can be. Mainly you have just to remove const
/// qualifiers from the variables you want to tune and flag them for tuning, so
/// if you have:
///
/// const Score myScore = S(10, 15);
/// const Value myValue[][2] = { { V(100), V(20) }, { V(7), V(78) } };
///
/// If you have a my_post_update() function to run after values have been updated,
/// and a my_range() function to set custom Option's min-max values, then you just
/// remove the 'const' qualifiers and write somewhere below in the file:
///
/// TUNE(SetRange(my_range), myScore, myValue, my_post_update);
///
/// You can also set the range directly, and restore the default at the end
///
/// TUNE(SetRange(-100, 100), myScore, SetDefaultRange);
///
/// In case update function is slow and you have many parameters, you can add:
///
/// UPDATE_ON_LAST();
///
/// And the values update, including post update function call, will be done only
/// once, after the engine receives the last UCI option, that is the one defined
/// and created as the last one, so the GUI should send the options in the same
/// order in which have been defined.
class Tune {
using PostUpdate = void (); // Post-update function
Tune() { read_results(); }
Tune(const Tune&) = delete;
void operator=(const Tune&) = delete;
void read_results();
static Tune& instance() { static Tune t; return t; } // Singleton
// Use polymorphism to accommodate Entry of different types in the same vector
struct EntryBase {
virtual ~EntryBase() = default;
virtual void init_option() = 0;
virtual void read_option() = 0;
};
template<typename T>
struct Entry : public EntryBase {
static_assert(!std::is_const<T>::value, "Parameter cannot be const!");
static_assert( std::is_same<T, int>::value
|| std::is_same<T, Value>::value
|| std::is_same<T, Score>::value
|| std::is_same<T, PostUpdate>::value, "Parameter type not supported!");
Entry(const std::string& n, T& v, const SetRange& r) : name(n), value(v), range(r) {}
void operator=(const Entry&) = delete; // Because 'value' is a reference
void init_option() override;
void read_option() override;
std::string name;
T& value;
SetRange range;
};
// Our facility to fill the container, each Entry corresponds to a parameter
// to tune. We use variadic templates to deal with an unspecified number of
// entries, each one of a possible different type.
static std::string next(std::string& names, bool pop = true);
int add(const SetRange&, std::string&&) { return 0; }
template<typename T, typename... Args>
int add(const SetRange& range, std::string&& names, T& value, Args&&... args) {
list.push_back(std::unique_ptr<EntryBase>(new Entry<T>(next(names), value, range)));
return add(range, std::move(names), args...);
}
// Template specialization for arrays: recursively handle multi-dimensional arrays
template<typename T, size_t N, typename... Args>
int add(const SetRange& range, std::string&& names, T (&value)[N], Args&&... args) {
for (size_t i = 0; i < N; i++)
add(range, next(names, i == N - 1) + "[" + std::to_string(i) + "]", value[i]);
return add(range, std::move(names), args...);
}
// Template specialization for SetRange
template<typename... Args>
int add(const SetRange&, std::string&& names, SetRange& value, Args&&... args) {
return add(value, (next(names), std::move(names)), args...);
}
std::vector<std::unique_ptr<EntryBase>> list;
public:
template<typename... Args>
static int add(const std::string& names, Args&&... args) {
return instance().add(SetDefaultRange, names.substr(1, names.size() - 2), args...); // Remove trailing parenthesis
}
static void init() { for (auto& e : instance().list) e->init_option(); read_options(); } // Deferred, due to UCI::Options access
static void read_options() { for (auto& e : instance().list) e->read_option(); }
static bool update_on_last;
};
// Some macro magic :-) we define a dummy int variable that compiler initializes calling Tune::add()
#define STRINGIFY(x) #x
#define UNIQUE2(x, y) x ## y
#define UNIQUE(x, y) UNIQUE2(x, y) // Two indirection levels to expand __LINE__
#define TUNE(...) int UNIQUE(p, __LINE__) = Tune::add(STRINGIFY((__VA_ARGS__)), __VA_ARGS__)
#define UPDATE_ON_LAST() bool UNIQUE(p, __LINE__) = Tune::update_on_last = true
} // namespace Stockfish
#endif // #ifndef TUNE_H_INCLUDED

491
src/Stockfish/src/types.h Normal file
View file

@ -0,0 +1,491 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TYPES_H_INCLUDED
#define TYPES_H_INCLUDED
/// When compiling with provided Makefile (e.g. for Linux and OSX), configuration
/// is done automatically. To get started type 'make help'.
///
/// When Makefile is not used (e.g. with Microsoft Visual Studio) some switches
/// need to be set manually:
///
/// -DNDEBUG | Disable debugging mode. Always use this for release.
///
/// -DNO_PREFETCH | Disable use of prefetch asm-instruction. You may need this to
/// | run on some very old machines.
///
/// -DUSE_POPCNT | Add runtime support for use of popcnt asm-instruction. Works
/// | only in 64-bit mode and requires hardware with popcnt support.
///
/// -DUSE_PEXT | Add runtime support for use of pext asm-instruction. Works
/// | only in 64-bit mode and requires hardware with pext support.
#include <cassert>
#include <cctype>
#include <cstdint>
#include <cstdlib>
#include <algorithm>
#if defined(_MSC_VER)
// Disable some silly and noisy warning from MSVC compiler
#pragma warning(disable: 4127) // Conditional expression is constant
#pragma warning(disable: 4146) // Unary minus operator applied to unsigned type
#pragma warning(disable: 4800) // Forcing value to bool 'true' or 'false'
#endif
/// Predefined macros hell:
///
/// __GNUC__ Compiler is gcc, Clang or Intel on Linux
/// __INTEL_COMPILER Compiler is Intel
/// _MSC_VER Compiler is MSVC or Intel on Windows
/// _WIN32 Building on Windows (any)
/// _WIN64 Building on Windows 64 bit
#if defined(__GNUC__ ) && (__GNUC__ < 9 || (__GNUC__ == 9 && __GNUC_MINOR__ <= 2)) && defined(_WIN32) && !defined(__clang__)
#define ALIGNAS_ON_STACK_VARIABLES_BROKEN
#endif
#define ASSERT_ALIGNED(ptr, alignment) assert(reinterpret_cast<uintptr_t>(ptr) % alignment == 0)
#if defined(_WIN64) && defined(_MSC_VER) // No Makefile used
# include <intrin.h> // Microsoft header for _BitScanForward64()
# define IS_64BIT
#endif
#if defined(USE_POPCNT) && (defined(__INTEL_COMPILER) || defined(_MSC_VER))
# include <nmmintrin.h> // Intel and Microsoft header for _mm_popcnt_u64()
#endif
#if !defined(NO_PREFETCH) && (defined(__INTEL_COMPILER) || defined(_MSC_VER))
# include <xmmintrin.h> // Intel and Microsoft header for _mm_prefetch()
#endif
#if defined(USE_PEXT)
# include <immintrin.h> // Header for _pext_u64() intrinsic
# define pext(b, m) _pext_u64(b, m)
#else
# define pext(b, m) 0
#endif
namespace Stockfish {
#ifdef USE_POPCNT
constexpr bool HasPopCnt = true;
#else
constexpr bool HasPopCnt = false;
#endif
#ifdef USE_PEXT
constexpr bool HasPext = true;
#else
constexpr bool HasPext = false;
#endif
#ifdef IS_64BIT
constexpr bool Is64Bit = true;
#else
constexpr bool Is64Bit = false;
#endif
using Key = uint64_t;
using Bitboard = uint64_t;
constexpr int MAX_MOVES = 256;
constexpr int MAX_PLY = 246;
/// A move needs 16 bits to be stored
///
/// bit 0- 5: destination square (from 0 to 63)
/// bit 6-11: origin square (from 0 to 63)
/// bit 12-13: promotion piece type - 2 (from KNIGHT-2 to QUEEN-2)
/// bit 14-15: special move flag: promotion (1), en passant (2), castling (3)
/// NOTE: en passant bit is set only when a pawn can be captured
///
/// Special cases are MOVE_NONE and MOVE_NULL. We can sneak these in because in
/// any normal move destination square is always different from origin square
/// while MOVE_NONE and MOVE_NULL have the same origin and destination square.
enum Move : int {
MOVE_NONE,
MOVE_NULL = 65
};
enum MoveType {
NORMAL,
PROMOTION = 1 << 14,
EN_PASSANT = 2 << 14,
CASTLING = 3 << 14
};
enum Color {
WHITE, BLACK, COLOR_NB = 2
};
enum CastlingRights {
NO_CASTLING,
WHITE_OO,
WHITE_OOO = WHITE_OO << 1,
BLACK_OO = WHITE_OO << 2,
BLACK_OOO = WHITE_OO << 3,
KING_SIDE = WHITE_OO | BLACK_OO,
QUEEN_SIDE = WHITE_OOO | BLACK_OOO,
WHITE_CASTLING = WHITE_OO | WHITE_OOO,
BLACK_CASTLING = BLACK_OO | BLACK_OOO,
ANY_CASTLING = WHITE_CASTLING | BLACK_CASTLING,
CASTLING_RIGHT_NB = 16
};
enum Phase {
PHASE_ENDGAME,
PHASE_MIDGAME = 128,
MG = 0, EG = 1, PHASE_NB = 2
};
enum ScaleFactor {
SCALE_FACTOR_DRAW = 0,
SCALE_FACTOR_NORMAL = 64,
SCALE_FACTOR_MAX = 128,
SCALE_FACTOR_NONE = 255
};
enum Bound {
BOUND_NONE,
BOUND_UPPER,
BOUND_LOWER,
BOUND_EXACT = BOUND_UPPER | BOUND_LOWER
};
enum Value : int {
VALUE_ZERO = 0,
VALUE_DRAW = 0,
VALUE_KNOWN_WIN = 10000,
VALUE_MATE = 32000,
VALUE_INFINITE = 32001,
VALUE_NONE = 32002,
VALUE_TB_WIN_IN_MAX_PLY = VALUE_MATE - 2 * MAX_PLY,
VALUE_TB_LOSS_IN_MAX_PLY = -VALUE_TB_WIN_IN_MAX_PLY,
VALUE_MATE_IN_MAX_PLY = VALUE_MATE - MAX_PLY,
VALUE_MATED_IN_MAX_PLY = -VALUE_MATE_IN_MAX_PLY,
// In the code, we make the assumption that these values
// are such that non_pawn_material() can be used to uniquely
// identify the material on the board.
PawnValueMg = 126, PawnValueEg = 208,
KnightValueMg = 781, KnightValueEg = 854,
BishopValueMg = 825, BishopValueEg = 915,
RookValueMg = 1276, RookValueEg = 1380,
QueenValueMg = 2538, QueenValueEg = 2682,
MidgameLimit = 15258, EndgameLimit = 3915
};
enum PieceType {
NO_PIECE_TYPE, PAWN, KNIGHT, BISHOP, ROOK, QUEEN, KING,
ALL_PIECES = 0,
PIECE_TYPE_NB = 8
};
enum Piece {
NO_PIECE,
W_PAWN = PAWN, W_KNIGHT, W_BISHOP, W_ROOK, W_QUEEN, W_KING,
B_PAWN = PAWN + 8, B_KNIGHT, B_BISHOP, B_ROOK, B_QUEEN, B_KING,
PIECE_NB = 16
};
constexpr Value PieceValue[PHASE_NB][PIECE_NB] = {
{ VALUE_ZERO, PawnValueMg, KnightValueMg, BishopValueMg, RookValueMg, QueenValueMg, VALUE_ZERO, VALUE_ZERO,
VALUE_ZERO, PawnValueMg, KnightValueMg, BishopValueMg, RookValueMg, QueenValueMg, VALUE_ZERO, VALUE_ZERO },
{ VALUE_ZERO, PawnValueEg, KnightValueEg, BishopValueEg, RookValueEg, QueenValueEg, VALUE_ZERO, VALUE_ZERO,
VALUE_ZERO, PawnValueEg, KnightValueEg, BishopValueEg, RookValueEg, QueenValueEg, VALUE_ZERO, VALUE_ZERO }
};
using Depth = int;
enum : int {
DEPTH_QS_CHECKS = 0,
DEPTH_QS_NO_CHECKS = -1,
DEPTH_QS_RECAPTURES = -5,
DEPTH_NONE = -6,
DEPTH_OFFSET = -7 // value used only for TT entry occupancy check
};
enum Square : int {
SQ_A1, SQ_B1, SQ_C1, SQ_D1, SQ_E1, SQ_F1, SQ_G1, SQ_H1,
SQ_A2, SQ_B2, SQ_C2, SQ_D2, SQ_E2, SQ_F2, SQ_G2, SQ_H2,
SQ_A3, SQ_B3, SQ_C3, SQ_D3, SQ_E3, SQ_F3, SQ_G3, SQ_H3,
SQ_A4, SQ_B4, SQ_C4, SQ_D4, SQ_E4, SQ_F4, SQ_G4, SQ_H4,
SQ_A5, SQ_B5, SQ_C5, SQ_D5, SQ_E5, SQ_F5, SQ_G5, SQ_H5,
SQ_A6, SQ_B6, SQ_C6, SQ_D6, SQ_E6, SQ_F6, SQ_G6, SQ_H6,
SQ_A7, SQ_B7, SQ_C7, SQ_D7, SQ_E7, SQ_F7, SQ_G7, SQ_H7,
SQ_A8, SQ_B8, SQ_C8, SQ_D8, SQ_E8, SQ_F8, SQ_G8, SQ_H8,
SQ_NONE,
SQUARE_ZERO = 0,
SQUARE_NB = 64
};
enum Direction : int {
NORTH = 8,
EAST = 1,
SOUTH = -NORTH,
WEST = -EAST,
NORTH_EAST = NORTH + EAST,
SOUTH_EAST = SOUTH + EAST,
SOUTH_WEST = SOUTH + WEST,
NORTH_WEST = NORTH + WEST
};
enum File : int {
FILE_A, FILE_B, FILE_C, FILE_D, FILE_E, FILE_F, FILE_G, FILE_H, FILE_NB
};
enum Rank : int {
RANK_1, RANK_2, RANK_3, RANK_4, RANK_5, RANK_6, RANK_7, RANK_8, RANK_NB
};
// Keep track of what a move changes on the board (used by NNUE)
struct DirtyPiece {
// Number of changed pieces
int dirty_num;
// Max 3 pieces can change in one move. A promotion with capture moves
// both the pawn and the captured piece to SQ_NONE and the piece promoted
// to from SQ_NONE to the capture square.
Piece piece[3];
// From and to squares, which may be SQ_NONE
Square from[3];
Square to[3];
};
/// Score enum stores a middlegame and an endgame value in a single integer (enum).
/// The least significant 16 bits are used to store the middlegame value and the
/// upper 16 bits are used to store the endgame value. We have to take care to
/// avoid left-shifting a signed int to avoid undefined behavior.
enum Score : int { SCORE_ZERO };
constexpr Score make_score(int mg, int eg) {
return Score((int)((unsigned int)eg << 16) + mg);
}
/// Extracting the signed lower and upper 16 bits is not so trivial because
/// according to the standard a simple cast to short is implementation defined
/// and so is a right shift of a signed integer.
inline Value eg_value(Score s) {
union { uint16_t u; int16_t s; } eg = { uint16_t(unsigned(s + 0x8000) >> 16) };
return Value(eg.s);
}
inline Value mg_value(Score s) {
union { uint16_t u; int16_t s; } mg = { uint16_t(unsigned(s)) };
return Value(mg.s);
}
#define ENABLE_BASE_OPERATORS_ON(T) \
constexpr T operator+(T d1, int d2) { return T(int(d1) + d2); } \
constexpr T operator-(T d1, int d2) { return T(int(d1) - d2); } \
constexpr T operator-(T d) { return T(-int(d)); } \
inline T& operator+=(T& d1, int d2) { return d1 = d1 + d2; } \
inline T& operator-=(T& d1, int d2) { return d1 = d1 - d2; }
#define ENABLE_INCR_OPERATORS_ON(T) \
inline T& operator++(T& d) { return d = T(int(d) + 1); } \
inline T& operator--(T& d) { return d = T(int(d) - 1); }
#define ENABLE_FULL_OPERATORS_ON(T) \
ENABLE_BASE_OPERATORS_ON(T) \
constexpr T operator*(int i, T d) { return T(i * int(d)); } \
constexpr T operator*(T d, int i) { return T(int(d) * i); } \
constexpr T operator/(T d, int i) { return T(int(d) / i); } \
constexpr int operator/(T d1, T d2) { return int(d1) / int(d2); } \
inline T& operator*=(T& d, int i) { return d = T(int(d) * i); } \
inline T& operator/=(T& d, int i) { return d = T(int(d) / i); }
ENABLE_FULL_OPERATORS_ON(Value)
ENABLE_FULL_OPERATORS_ON(Direction)
ENABLE_INCR_OPERATORS_ON(Piece)
ENABLE_INCR_OPERATORS_ON(PieceType)
ENABLE_INCR_OPERATORS_ON(Square)
ENABLE_INCR_OPERATORS_ON(File)
ENABLE_INCR_OPERATORS_ON(Rank)
ENABLE_BASE_OPERATORS_ON(Score)
#undef ENABLE_FULL_OPERATORS_ON
#undef ENABLE_INCR_OPERATORS_ON
#undef ENABLE_BASE_OPERATORS_ON
/// Additional operators to add a Direction to a Square
constexpr Square operator+(Square s, Direction d) { return Square(int(s) + int(d)); }
constexpr Square operator-(Square s, Direction d) { return Square(int(s) - int(d)); }
inline Square& operator+=(Square& s, Direction d) { return s = s + d; }
inline Square& operator-=(Square& s, Direction d) { return s = s - d; }
/// Only declared but not defined. We don't want to multiply two scores due to
/// a very high risk of overflow. So user should explicitly convert to integer.
Score operator*(Score, Score) = delete;
/// Division of a Score must be handled separately for each term
inline Score operator/(Score s, int i) {
return make_score(mg_value(s) / i, eg_value(s) / i);
}
/// Multiplication of a Score by an integer. We check for overflow in debug mode.
inline Score operator*(Score s, int i) {
Score result = Score(int(s) * i);
assert(eg_value(result) == (i * eg_value(s)));
assert(mg_value(result) == (i * mg_value(s)));
assert((i == 0) || (result / i) == s);
return result;
}
/// Multiplication of a Score by a boolean
inline Score operator*(Score s, bool b) {
return b ? s : SCORE_ZERO;
}
constexpr Color operator~(Color c) {
return Color(c ^ BLACK); // Toggle color
}
constexpr Square flip_rank(Square s) { // Swap A1 <-> A8
return Square(s ^ SQ_A8);
}
constexpr Square flip_file(Square s) { // Swap A1 <-> H1
return Square(s ^ SQ_H1);
}
constexpr Piece operator~(Piece pc) {
return Piece(pc ^ 8); // Swap color of piece B_KNIGHT <-> W_KNIGHT
}
constexpr CastlingRights operator&(Color c, CastlingRights cr) {
return CastlingRights((c == WHITE ? WHITE_CASTLING : BLACK_CASTLING) & cr);
}
constexpr Value mate_in(int ply) {
return VALUE_MATE - ply;
}
constexpr Value mated_in(int ply) {
return -VALUE_MATE + ply;
}
constexpr Square make_square(File f, Rank r) {
return Square((r << 3) + f);
}
constexpr Piece make_piece(Color c, PieceType pt) {
return Piece((c << 3) + pt);
}
constexpr PieceType type_of(Piece pc) {
return PieceType(pc & 7);
}
inline Color color_of(Piece pc) {
assert(pc != NO_PIECE);
return Color(pc >> 3);
}
constexpr bool is_ok(Move m) {
return m != MOVE_NONE && m != MOVE_NULL;
}
constexpr bool is_ok(Square s) {
return s >= SQ_A1 && s <= SQ_H8;
}
constexpr File file_of(Square s) {
return File(s & 7);
}
constexpr Rank rank_of(Square s) {
return Rank(s >> 3);
}
constexpr Square relative_square(Color c, Square s) {
return Square(s ^ (c * 56));
}
constexpr Rank relative_rank(Color c, Rank r) {
return Rank(r ^ (c * 7));
}
constexpr Rank relative_rank(Color c, Square s) {
return relative_rank(c, rank_of(s));
}
constexpr Direction pawn_push(Color c) {
return c == WHITE ? NORTH : SOUTH;
}
constexpr Square from_sq(Move m) {
assert(is_ok(m));
return Square((m >> 6) & 0x3F);
}
constexpr Square to_sq(Move m) {
assert(is_ok(m));
return Square(m & 0x3F);
}
constexpr int from_to(Move m) {
return m & 0xFFF;
}
constexpr MoveType type_of(Move m) {
return MoveType(m & (3 << 14));
}
constexpr PieceType promotion_type(Move m) {
return PieceType(((m >> 12) & 3) + KNIGHT);
}
constexpr Move make_move(Square from, Square to) {
return Move((from << 6) + to);
}
template<MoveType T>
constexpr Move make(Square from, Square to, PieceType pt = KNIGHT) {
return Move(T + ((pt - KNIGHT) << 12) + (from << 6) + to);
}
/// Based on a congruential pseudo random number generator
constexpr Key make_key(uint64_t seed) {
return seed * 6364136223846793005ULL + 1442695040888963407ULL;
}
} // namespace Stockfish
#endif // #ifndef TYPES_H_INCLUDED
#include "tune.h" // Global visibility to tuning setup

398
src/Stockfish/src/uci.cpp Normal file
View file

@ -0,0 +1,398 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <cassert>
#include <cmath>
#include <iostream>
#include <sstream>
#include <string>
#include "benchmark.h"
#include "evaluate.h"
#include "movegen.h"
#include "position.h"
#include "search.h"
#include "thread.h"
#include "timeman.h"
#include "tt.h"
#include "uci.h"
#include "syzygy/tbprobe.h"
#include "nnue/evaluate_nnue.h"
using namespace std;
namespace Stockfish {
namespace {
// FEN string for the initial position in standard chess
const char* StartFEN = "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1";
// position() is called when the engine receives the "position" UCI command.
// It sets up the position that is described in the given FEN string ("fen") or
// the initial position ("startpos") and then makes the moves given in the following
// move list ("moves").
void position(Position& pos, istringstream& is, StateListPtr& states) {
Move m;
string token, fen;
is >> token;
if (token == "startpos")
{
fen = StartFEN;
is >> token; // Consume the "moves" token, if any
}
else if (token == "fen")
while (is >> token && token != "moves")
fen += token + " ";
else
return;
states = StateListPtr(new std::deque<StateInfo>(1)); // Drop the old state and create a new one
pos.set(fen, Options["UCI_Chess960"], &states->back(), Threads.main());
// Parse the move list, if any
while (is >> token && (m = UCI::to_move(pos, token)) != MOVE_NONE)
{
states->emplace_back();
pos.do_move(m, states->back());
}
}
// trace_eval() prints the evaluation of the current position, consistent with
// the UCI options set so far.
void trace_eval(Position& pos) {
StateListPtr states(new std::deque<StateInfo>(1));
Position p;
p.set(pos.fen(), Options["UCI_Chess960"], &states->back(), Threads.main());
Eval::NNUE::verify();
sync_cout << "\n" << Eval::trace(p) << sync_endl;
}
// setoption() is called when the engine receives the "setoption" UCI command.
// The function updates the UCI option ("name") to the given value ("value").
void setoption(istringstream& is) {
string token, name, value;
is >> token; // Consume the "name" token
// Read the option name (can contain spaces)
while (is >> token && token != "value")
name += (name.empty() ? "" : " ") + token;
// Read the option value (can contain spaces)
while (is >> token)
value += (value.empty() ? "" : " ") + token;
if (Options.count(name))
Options[name] = value;
else
sync_cout << "No such option: " << name << sync_endl;
}
// go() is called when the engine receives the "go" UCI command. The function
// sets the thinking time and other parameters from the input string, then starts
// with a search.
void go(Position& pos, istringstream& is, StateListPtr& states) {
Search::LimitsType limits;
string token;
bool ponderMode = false;
limits.startTime = now(); // The search starts as early as possible
while (is >> token)
if (token == "searchmoves") // Needs to be the last command on the line
while (is >> token)
limits.searchmoves.push_back(UCI::to_move(pos, token));
else if (token == "wtime") is >> limits.time[WHITE];
else if (token == "btime") is >> limits.time[BLACK];
else if (token == "winc") is >> limits.inc[WHITE];
else if (token == "binc") is >> limits.inc[BLACK];
else if (token == "movestogo") is >> limits.movestogo;
else if (token == "depth") is >> limits.depth;
else if (token == "nodes") is >> limits.nodes;
else if (token == "movetime") is >> limits.movetime;
else if (token == "mate") is >> limits.mate;
else if (token == "perft") is >> limits.perft;
else if (token == "infinite") limits.infinite = 1;
else if (token == "ponder") ponderMode = true;
Threads.start_thinking(pos, states, limits, ponderMode);
}
// bench() is called when the engine receives the "bench" command.
// Firstly, a list of UCI commands is set up according to the bench
// parameters, then it is run one by one, printing a summary at the end.
void bench(Position& pos, istream& args, StateListPtr& states) {
string token;
uint64_t num, nodes = 0, cnt = 1;
vector<string> list = setup_bench(pos, args);
num = count_if(list.begin(), list.end(), [](const string& s) { return s.find("go ") == 0 || s.find("eval") == 0; });
TimePoint elapsed = now();
for (const auto& cmd : list)
{
istringstream is(cmd);
is >> skipws >> token;
if (token == "go" || token == "eval")
{
cerr << "\nPosition: " << cnt++ << '/' << num << " (" << pos.fen() << ")" << endl;
if (token == "go")
{
go(pos, is, states);
Threads.main()->wait_for_search_finished();
nodes += Threads.nodes_searched();
}
else
trace_eval(pos);
}
else if (token == "setoption") setoption(is);
else if (token == "position") position(pos, is, states);
else if (token == "ucinewgame") { Search::clear(); elapsed = now(); } // Search::clear() may take a while
}
elapsed = now() - elapsed + 1; // Ensure positivity to avoid a 'divide by zero'
dbg_print();
cerr << "\n==========================="
<< "\nTotal time (ms) : " << elapsed
<< "\nNodes searched : " << nodes
<< "\nNodes/second : " << 1000 * nodes / elapsed << endl;
}
// The win rate model returns the probability of winning (in per mille units) given an
// eval and a game ply. It fits the LTC fishtest statistics rather accurately.
int win_rate_model(Value v, int ply) {
// The model only captures up to 240 plies, so limit the input and then rescale
double m = std::min(240, ply) / 64.0;
// The coefficients of a third-order polynomial fit is based on the fishtest data
// for two parameters that need to transform eval to the argument of a logistic
// function.
constexpr double as[] = { 0.38036525, -2.82015070, 23.17882135, 307.36768407};
constexpr double bs[] = { -2.29434733, 13.27689788, -14.26828904, 63.45318330 };
// Enforce that NormalizeToPawnValue corresponds to a 50% win rate at ply 64
static_assert(UCI::NormalizeToPawnValue == int(as[0] + as[1] + as[2] + as[3]));
double a = (((as[0] * m + as[1]) * m + as[2]) * m) + as[3];
double b = (((bs[0] * m + bs[1]) * m + bs[2]) * m) + bs[3];
// Transform the eval to centipawns with limited range
double x = std::clamp(double(v), -4000.0, 4000.0);
// Return the win rate in per mille units rounded to the nearest value
return int(0.5 + 1000 / (1 + std::exp((a - x) / b)));
}
} // namespace
/// UCI::loop() waits for a command from the stdin, parses it and then calls the appropriate
/// function. It also intercepts an end-of-file (EOF) indication from the stdin to ensure a
/// graceful exit if the GUI dies unexpectedly. When called with some command-line arguments,
/// like running 'bench', the function returns immediately after the command is executed.
/// In addition to the UCI ones, some additional debug commands are also supported.
void UCI::loop(int argc, char* argv[]) {
Position pos;
string token, cmd;
StateListPtr states(new std::deque<StateInfo>(1));
pos.set(StartFEN, false, &states->back(), Threads.main());
for (int i = 1; i < argc; ++i)
cmd += std::string(argv[i]) + " ";
do {
if (argc == 1 && !getline(fakein, cmd)) // Wait for an input or an end-of-file (EOF) indication
cmd = "quit";
istringstream is(cmd);
token.clear(); // Avoid a stale if getline() returns nothing or a blank line
is >> skipws >> token;
if ( token == "quit"
|| token == "stop")
Threads.stop = true;
// The GUI sends 'ponderhit' to tell that the user has played the expected move.
// So, 'ponderhit' is sent if pondering was done on the same move that the user
// has played. The search should continue, but should also switch from pondering
// to the normal search.
else if (token == "ponderhit")
Threads.main()->ponder = false; // Switch to the normal search
else if (token == "uci")
sync_cout << "id name " << engine_info(true)
<< "\n" << Options
<< "\nuciok" << sync_endl;
else if (token == "setoption") setoption(is);
else if (token == "go") go(pos, is, states);
else if (token == "position") position(pos, is, states);
else if (token == "ucinewgame") Search::clear();
else if (token == "isready") sync_cout << "readyok" << sync_endl;
// Add custom non-UCI commands, mainly for debugging purposes.
// These commands must not be used during a search!
else if (token == "flip") pos.flip();
else if (token == "bench") bench(pos, is, states);
else if (token == "d") sync_cout << pos << sync_endl;
else if (token == "eval") trace_eval(pos);
else if (token == "compiler") sync_cout << compiler_info() << sync_endl;
else if (token == "export_net")
{
std::optional<std::string> filename;
std::string f;
if (is >> skipws >> f)
filename = f;
Eval::NNUE::save_eval(filename);
}
else if (token == "--help" || token == "help" || token == "--license" || token == "license")
sync_cout << "\nStockfish is a powerful chess engine for playing and analyzing."
"\nIt is released as free software licensed under the GNU GPLv3 License."
"\nStockfish is normally used with a graphical user interface (GUI) and implements"
"\nthe Universal Chess Interface (UCI) protocol to communicate with a GUI, an API, etc."
"\nFor any further information, visit https://github.com/official-stockfish/Stockfish#readme"
"\nor read the corresponding README.md and Copying.txt files distributed along with this program.\n" << sync_endl;
else if (!token.empty() && token[0] != '#')
sync_cout << "Unknown command: '" << cmd << "'. Type help for more information." << sync_endl;
} while (token != "quit" && argc == 1); // The command-line arguments are one-shot
}
/// UCI::value() converts a Value to a string by adhering to the UCI protocol specification:
///
/// cp <x> The score from the engine's point of view in centipawns.
/// mate <y> Mate in 'y' moves (not plies). If the engine is getting mated,
/// uses negative values for 'y'.
string UCI::value(Value v) {
assert(-VALUE_INFINITE < v && v < VALUE_INFINITE);
stringstream ss;
if (abs(v) < VALUE_TB_WIN_IN_MAX_PLY)
ss << "cp " << v * 100 / NormalizeToPawnValue;
else if (abs(v) < VALUE_MATE_IN_MAX_PLY)
{
const int ply = VALUE_MATE_IN_MAX_PLY - 1 - std::abs(v); // recompute ss->ply
ss << "cp " << (v > 0 ? 20000 - ply : -20000 + ply);
}
else
ss << "mate " << (v > 0 ? VALUE_MATE - v + 1 : -VALUE_MATE - v) / 2;
return ss.str();
}
/// UCI::wdl() reports the win-draw-loss (WDL) statistics given an evaluation
/// and a game ply based on the data gathered for fishtest LTC games.
string UCI::wdl(Value v, int ply) {
stringstream ss;
int wdl_w = win_rate_model( v, ply);
int wdl_l = win_rate_model(-v, ply);
int wdl_d = 1000 - wdl_w - wdl_l;
ss << " wdl " << wdl_w << " " << wdl_d << " " << wdl_l;
return ss.str();
}
/// UCI::square() converts a Square to a string in algebraic notation (g1, a7, etc.)
std::string UCI::square(Square s) {
return std::string{ char('a' + file_of(s)), char('1' + rank_of(s)) };
}
/// UCI::move() converts a Move to a string in coordinate notation (g1f3, a7a8q).
/// The only special case is castling where the e1g1 notation is printed in
/// standard chess mode and in e1h1 notation it is printed in Chess960 mode.
/// Internally, all castling moves are always encoded as 'king captures rook'.
string UCI::move(Move m, bool chess960) {
if (m == MOVE_NONE)
return "(none)";
if (m == MOVE_NULL)
return "0000";
Square from = from_sq(m);
Square to = to_sq(m);
if (type_of(m) == CASTLING && !chess960)
to = make_square(to > from ? FILE_G : FILE_C, rank_of(from));
string move = UCI::square(from) + UCI::square(to);
if (type_of(m) == PROMOTION)
move += " pnbrqk"[promotion_type(m)];
return move;
}
/// UCI::to_move() converts a string representing a move in coordinate notation
/// (g1f3, a7a8q) to the corresponding legal Move, if any.
Move UCI::to_move(const Position& pos, string& str) {
if (str.length() == 5)
str[4] = char(tolower(str[4])); // The promotion piece character must be lowercased
for (const auto& m : MoveList<LEGAL>(pos))
if (str == UCI::move(m, pos.is_chess960()))
return m;
return MOVE_NONE;
}
} // namespace Stockfish

92
src/Stockfish/src/uci.h Normal file
View file

@ -0,0 +1,92 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef UCI_H_INCLUDED
#define UCI_H_INCLUDED
#include <map>
#include <string>
#include "types.h"
namespace Stockfish {
class Position;
namespace UCI {
// Normalizes the internal value as reported by evaluate or search
// to the UCI centipawn result used in output. This value is derived from
// the win_rate_model() such that Stockfish outputs an advantage of
// "100 centipawns" for a position if the engine has a 50% probability to win
// from this position in selfplay at fishtest LTC time control.
const int NormalizeToPawnValue = 328;
class Option;
/// Define a custom comparator, because the UCI options should be case-insensitive
struct CaseInsensitiveLess {
bool operator() (const std::string&, const std::string&) const;
};
/// The options container is defined as a std::map
using OptionsMap = std::map<std::string, Option, CaseInsensitiveLess>;
/// The Option class implements each option as specified by the UCI protocol
class Option {
using OnChange = void (*)(const Option&);
public:
Option(OnChange = nullptr);
Option(bool v, OnChange = nullptr);
Option(const char* v, OnChange = nullptr);
Option(double v, int minv, int maxv, OnChange = nullptr);
Option(const char* v, const char* cur, OnChange = nullptr);
Option& operator=(const std::string&);
void operator<<(const Option&);
operator int() const;
operator std::string() const;
bool operator==(const char*) const;
private:
friend std::ostream& operator<<(std::ostream&, const OptionsMap&);
std::string defaultValue, currentValue, type;
int min, max;
size_t idx;
OnChange on_change;
};
void init(OptionsMap&);
void loop(int argc, char* argv[]);
std::string value(Value v);
std::string square(Square s);
std::string move(Move m, bool chess960);
std::string pv(const Position& pos, Depth depth);
std::string wdl(Value v, int ply);
Move to_move(const Position& pos, std::string& str);
} // namespace UCI
extern UCI::OptionsMap Options;
} // namespace Stockfish
#endif // #ifndef UCI_H_INCLUDED

194
src/Stockfish/src/ucioption.cpp Normal file
View file

@ -0,0 +1,194 @@
/*
Stockfish, a UCI chess playing engine derived from Glaurung 2.1
Copyright (C) 2004-2023 The Stockfish developers (see AUTHORS file)
Stockfish is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Stockfish is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <algorithm>
#include <cassert>
#include <ostream>
#include <sstream>
#include "evaluate.h"
#include "misc.h"
#include "search.h"
#include "thread.h"
#include "tt.h"
#include "uci.h"
#include "syzygy/tbprobe.h"
using std::string;
namespace Stockfish {
UCI::OptionsMap Options; // Global object
namespace UCI {
/// 'On change' actions, triggered by an option's value change
static void on_clear_hash(const Option&) { Search::clear(); }
static void on_hash_size(const Option& o) { TT.resize(size_t(o)); }
static void on_logger(const Option& o) { start_logger(o); }
static void on_threads(const Option& o) { Threads.set(size_t(o)); }
static void on_tb_path(const Option& o) { Tablebases::init(o); }
static void on_use_NNUE(const Option&) { Eval::NNUE::init(); }
static void on_eval_file(const Option&) { Eval::NNUE::init(); }
/// Our case insensitive less() function as required by UCI protocol
bool CaseInsensitiveLess::operator() (const string& s1, const string& s2) const {
return std::lexicographical_compare(s1.begin(), s1.end(), s2.begin(), s2.end(),
[](char c1, char c2) { return tolower(c1) < tolower(c2); });
}
/// UCI::init() initializes the UCI options to their hard-coded default values
void init(OptionsMap& o) {
constexpr int MaxHashMB = Is64Bit ? 33554432 : 2048;
o["Debug Log File"] << Option("", on_logger);
o["Threads"] << Option(1, 1, 1024, on_threads);
o["Hash"] << Option(16, 1, MaxHashMB, on_hash_size);
o["Clear Hash"] << Option(on_clear_hash);
o["Ponder"] << Option(false);
o["MultiPV"] << Option(1, 1, 500);
o["Skill Level"] << Option(20, 0, 20);
o["Move Overhead"] << Option(10, 0, 5000);
o["Slow Mover"] << Option(100, 10, 1000);
o["nodestime"] << Option(0, 0, 10000);
o["UCI_Chess960"] << Option(false);
o["UCI_AnalyseMode"] << Option(false);
o["UCI_LimitStrength"] << Option(false);
o["UCI_Elo"] << Option(1320, 1320, 3190);
o["UCI_ShowWDL"] << Option(false);
o["SyzygyPath"] << Option("<empty>", on_tb_path);
o["SyzygyProbeDepth"] << Option(1, 1, 100);
o["Syzygy50MoveRule"] << Option(true);
o["SyzygyProbeLimit"] << Option(7, 0, 7);
o["Use NNUE"] << Option(true, on_use_NNUE);
o["EvalFile"] << Option(EvalFileDefaultName, on_eval_file);
}
/// operator<<() is used to print all the options default values in chronological
/// insertion order (the idx field) and in the format defined by the UCI protocol.
std::ostream& operator<<(std::ostream& os, const OptionsMap& om) {
for (size_t idx = 0; idx < om.size(); ++idx)
for (const auto& it : om)
if (it.second.idx == idx)
{
const Option& o = it.second;
os << "\noption name " << it.first << " type " << o.type;
if (o.type == "string" || o.type == "check" || o.type == "combo")
os << " default " << o.defaultValue;
if (o.type == "spin")
os << " default " << int(stof(o.defaultValue))
<< " min " << o.min
<< " max " << o.max;
break;
}
return os;
}
/// Option class constructors and conversion operators
Option::Option(const char* v, OnChange f) : type("string"), min(0), max(0), on_change(f)
{ defaultValue = currentValue = v; }
Option::Option(bool v, OnChange f) : type("check"), min(0), max(0), on_change(f)
{ defaultValue = currentValue = (v ? "true" : "false"); }
Option::Option(OnChange f) : type("button"), min(0), max(0), on_change(f)
{}
Option::Option(double v, int minv, int maxv, OnChange f) : type("spin"), min(minv), max(maxv), on_change(f)
{ defaultValue = currentValue = std::to_string(v); }
Option::Option(const char* v, const char* cur, OnChange f) : type("combo"), min(0), max(0), on_change(f)
{ defaultValue = v; currentValue = cur; }
Option::operator int() const {
assert(type == "check" || type == "spin");
return (type == "spin" ? std::stoi(currentValue) : currentValue == "true");
}
Option::operator std::string() const {
assert(type == "string");
return currentValue;
}
bool Option::operator==(const char* s) const {
assert(type == "combo");
return !CaseInsensitiveLess()(currentValue, s)
&& !CaseInsensitiveLess()(s, currentValue);
}
/// operator<<() inits options and assigns idx in the correct printing order
void Option::operator<<(const Option& o) {
static size_t insert_order = 0;
*this = o;
idx = insert_order++;
}
/// operator=() updates currentValue and triggers on_change() action. It's up to
/// the GUI to check for option's limits, but we could receive the new value
/// from the user by console window, so let's check the bounds anyway.
Option& Option::operator=(const string& v) {
assert(!type.empty());
if ( (type != "button" && type != "string" && v.empty())
|| (type == "check" && v != "true" && v != "false")
|| (type == "spin" && (stof(v) < min || stof(v) > max)))
return *this;
if (type == "combo")
{
OptionsMap comboMap; // To have case insensitive compare
string token;
std::istringstream ss(defaultValue);
while (ss >> token)
comboMap[token] << Option();
if (!comboMap.count(v) || v == "var")
return *this;
}
if (type != "button")
currentValue = v;
if (on_change)
on_change(*this);
return *this;
}
} // namespace UCI
} // namespace Stockfish

151
src/Stockfish/tests/instrumented.sh Executable file
View file

@ -0,0 +1,151 @@
#!/bin/bash
# check for errors under valgrind or sanitizers.
error()
{
echo "instrumented testing failed on line $1"
exit 1
}
trap 'error ${LINENO}' ERR
# define suitable post and prefixes for testing options
case $1 in
--valgrind)
echo "valgrind testing started"
prefix=''
exeprefix='valgrind --error-exitcode=42 --errors-for-leak-kinds=all --leak-check=full'
postfix='1>/dev/null'
threads="1"
;;
--valgrind-thread)
echo "valgrind-thread testing started"
prefix=''
exeprefix='valgrind --fair-sched=try --error-exitcode=42'
postfix='1>/dev/null'
threads="2"
;;
--sanitizer-undefined)
echo "sanitizer-undefined testing started"
prefix='!'
exeprefix=''
postfix='2>&1 | grep -A50 "runtime error:"'
threads="1"
;;
--sanitizer-thread)
echo "sanitizer-thread testing started"
prefix='!'
exeprefix=''
postfix='2>&1 | grep -A50 "WARNING: ThreadSanitizer:"'
threads="2"
cat << EOF > tsan.supp
race:Stockfish::TTEntry::move
race:Stockfish::TTEntry::depth
race:Stockfish::TTEntry::bound
race:Stockfish::TTEntry::save
race:Stockfish::TTEntry::value
race:Stockfish::TTEntry::eval
race:Stockfish::TTEntry::is_pv
race:Stockfish::TranspositionTable::probe
race:Stockfish::TranspositionTable::hashfull
EOF
export TSAN_OPTIONS="suppressions=./tsan.supp"
;;
*)
echo "unknown testing started"
prefix=''
exeprefix=''
postfix=''
threads="1"
;;
esac
# simple command line testing
for args in "eval" \
"go nodes 1000" \
"go depth 10" \
"go movetime 1000" \
"go wtime 8000 btime 8000 winc 500 binc 500" \
"bench 128 $threads 8 default depth" \
"export_net verify.nnue"
do
echo "$prefix $exeprefix ./stockfish $args $postfix"
eval "$prefix $exeprefix ./stockfish $args $postfix"
done
# verify the generated net equals the base net
network=`./stockfish uci | grep 'option name EvalFile type string default' | awk '{print $NF}'`
echo "Comparing $network to the written verify.nnue"
diff $network verify.nnue
# more general testing, following an uci protocol exchange
cat << EOF > game.exp
set timeout 240
spawn $exeprefix ./stockfish
send "uci\n"
expect "uciok"
send "setoption name Threads value $threads\n"
send "ucinewgame\n"
send "position startpos\n"
send "go nodes 1000\n"
expect "bestmove"
send "position startpos moves e2e4 e7e6\n"
send "go nodes 1000\n"
expect "bestmove"
send "position fen 5rk1/1K4p1/8/8/3B4/8/8/8 b - - 0 1\n"
send "go depth 10\n"
expect "bestmove"
send "quit\n"
expect eof
# return error code of the spawned program, useful for valgrind
lassign [wait] pid spawnid os_error_flag value
exit \$value
EOF
#download TB as needed
if [ ! -d ../tests/syzygy ]; then
curl -sL https://api.github.com/repos/niklasf/python-chess/tarball/9b9aa13f9f36d08aadfabff872882f4ab1494e95 | tar -xzf -
mv niklasf-python-chess-9b9aa13 ../tests/syzygy
fi
cat << EOF > syzygy.exp
set timeout 240
spawn $exeprefix ./stockfish
send "uci\n"
send "setoption name SyzygyPath value ../tests/syzygy/\n"
expect "info string Found 35 tablebases" {} timeout {exit 1}
send "bench 128 1 8 default depth\n"
send "quit\n"
expect eof
# return error code of the spawned program, useful for valgrind
lassign [wait] pid spawnid os_error_flag value
exit \$value
EOF
for exp in game.exp syzygy.exp
do
echo "$prefix expect $exp $postfix"
eval "$prefix expect $exp $postfix"
rm $exp
done
rm -f tsan.supp
echo "instrumented testing OK"

32
src/Stockfish/tests/perft.sh Executable file
View file

@ -0,0 +1,32 @@
#!/bin/bash
# verify perft numbers (positions from www.chessprogramming.org/Perft_Results)
error()
{
echo "perft testing failed on line $1"
exit 1
}
trap 'error ${LINENO}' ERR
echo "perft testing started"
cat << EOF > perft.exp
set timeout 10
lassign \$argv pos depth result
spawn ./stockfish
send "position \$pos\\ngo perft \$depth\\n"
expect "Nodes searched? \$result" {} timeout {exit 1}
send "quit\\n"
expect eof
EOF
expect perft.exp startpos 5 4865609 > /dev/null
expect perft.exp "fen r3k2r/p1ppqpb1/bn2pnp1/3PN3/1p2P3/2N2Q1p/PPPBBPPP/R3K2R w KQkq -" 5 193690690 > /dev/null
expect perft.exp "fen 8/2p5/3p4/KP5r/1R3p1k/8/4P1P1/8 w - -" 6 11030083 > /dev/null
expect perft.exp "fen r3k2r/Pppp1ppp/1b3nbN/nP6/BBP1P3/q4N2/Pp1P2PP/R2Q1RK1 w kq - 0 1" 5 15833292 > /dev/null
expect perft.exp "fen rnbq1k1r/pp1Pbppp/2p5/8/2B5/8/PPP1NnPP/RNBQK2R w KQ - 1 8" 5 89941194 > /dev/null
expect perft.exp "fen r4rk1/1pp1qppp/p1np1n2/2b1p1B1/2B1P1b1/P1NP1N2/1PP1QPPP/R4RK1 w - - 0 10" 5 164075551 > /dev/null
rm perft.exp
echo "perft testing OK"

61
src/Stockfish/tests/reprosearch.sh Executable file
View file

@ -0,0 +1,61 @@
#!/bin/bash
# verify reproducible search
error()
{
echo "reprosearch testing failed on line $1"
exit 1
}
trap 'error ${LINENO}' ERR
echo "reprosearch testing started"
# repeat two short games, separated by ucinewgame.
# with go nodes $nodes they should result in exactly
# the same node count for each iteration.
cat << EOF > repeat.exp
set timeout 10
spawn ./stockfish
lassign \$argv nodes
send "uci\n"
expect "uciok"
send "ucinewgame\n"
send "position startpos\n"
send "go nodes \$nodes\n"
expect "bestmove"
send "position startpos moves e2e4 e7e6\n"
send "go nodes \$nodes\n"
expect "bestmove"
send "ucinewgame\n"
send "position startpos\n"
send "go nodes \$nodes\n"
expect "bestmove"
send "position startpos moves e2e4 e7e6\n"
send "go nodes \$nodes\n"
expect "bestmove"
send "quit\n"
expect eof
EOF
# to increase the likelihood of finding a non-reproducible case,
# the allowed number of nodes are varied systematically
for i in `seq 1 20`
do
nodes=$((100*3**i/2**i))
echo "reprosearch testing with $nodes nodes"
# each line should appear exactly an even number of times
expect repeat.exp $nodes 2>&1 | grep -o "nodes [0-9]*" | sort | uniq -c | awk '{if ($1%2!=0) exit(1)}'
done
rm repeat.exp
echo "reprosearch testing OK"

31
src/Stockfish/tests/signature.sh Executable file
View file

@ -0,0 +1,31 @@
#!/bin/bash
# obtain and optionally verify Bench / signature
# if no reference is given, the output is deliberately limited to just the signature
error()
{
echo "running bench for signature failed on line $1"
exit 1
}
trap 'error ${LINENO}' ERR
# obtain
signature=`./stockfish bench 2>&1 | grep "Nodes searched : " | awk '{print $4}'`
if [ $# -gt 0 ]; then
# compare to given reference
if [ "$1" != "$signature" ]; then
if [ -z "$signature" ]; then
echo "No signature obtained from bench. Code crashed or assert triggered ?"
else
echo "signature mismatch: reference $1 obtained: $signature ."
fi
exit 1
else
echo "signature OK: $signature"
fi
else
# just report signature
echo $signature
fi

21
src/create_stockfish.sh Normal file
View file

@ -0,0 +1,21 @@
curl -s https://api.github.com/repos/official-stockfish/Stockfish/releases/latest \
| grep "tarball_url" \
| cut -d : -f 2,3 \
| tr -d "\", " \
| wget -qi - -O Stockfish.tar.gz
rm -rf Stockfish
mkdir -p Stockfish
tar -xzf Stockfish.tar.gz --directory Stockfish --strip-components=1
rm Stockfish.tar.gz
find ./Stockfish/ -type f -exec sed -i \
-e 's/std::cout/fakeout/g' \
-e 's/std::cin/fakein/g' \
-e 's/std::endl/fakeendl/g' \
-e 's/getline(cin, cmd)/getline(fakein, cmd)/g' \
{} +
nnue_name=$(grep EvalFileDefaultName Stockfish/src/evaluate.h \
| grep define \
| sed 's/.*\(nn-[a-z0-9]\{12\}.nnue\).*/\1/')
sed -i "s/set(NNUE_NAME nn-[^)]*.nnue)/set(NNUE_NAME $nnue_name)/" CMakeLists.txt

2
src/stream_fix.cpp
View file

@ -17,7 +17,7 @@ bool FakeStream::is_closed() { return closed; }
std::streambuf* FakeStream::rdbuf() { return nullptr; }
std::streambuf* FakeStream::rdbuf(std::streambuf* __sb) { return nullptr; }
std::streambuf* FakeStream::rdbuf(std::streambuf* buf) { return nullptr; }
bool std::getline(FakeStream& is, std::string& str) {
if (is.is_closed()) return false;

17
windows/.gitignore vendored Normal file
View file

@ -0,0 +1,17 @@
flutter/
# Visual Studio user-specific files.
*.suo
*.user
*.userosscache
*.sln.docstates
# Visual Studio build-related files.
x64/
x86/
# Visual Studio cache files
# files ending in .cache can be ignored
*.[Cc]ache
# but keep track of directories ending in .cache
!*.[Cc]ache/