Stockfish

История

Визуализация NNUE Stockfish

Программа возникла на основе Glaurung , шахматного движка с открытым исходным кодом, созданного Ромстадом и впервые выпущенного в 2004 году. Четыре года спустя Костальба, вдохновленный сильным движком с открытым исходным кодом, решил разделить проект. Он назвал его Stockfish, потому что он был «произведен в Норвегии и приготовлен в Италии» (Romstad — норвежский, Costalba — итальянский). Первая версия Stockfish 1.0 была выпущена в ноябре 2008 года. Некоторое время новые идеи и изменения кода передавались между двумя программами в обоих направлениях, пока Ромстад не решил прекратить выпуск Glaurung в пользу Stockfish, который был более продвинутым движком. в то время. Последний Glaurung (версия 2.2) был выпущен в декабре 2008 года.

Примерно в 2011 году Ромстад решил отказаться от участия в Stockfish, чтобы больше времени уделять своему новому шахматному приложению для iOS.

18 июня 2014 года Марко Костальба объявил, что «решил уйти с должности сопровождающего Stockfish», и попросил сообщество создать форк текущей версии и продолжить ее разработку. Официальный репозиторий, которым управляет группа добровольцев основных разработчиков Stockfish, был создан вскоре после этого и в настоящее время управляет развитием проекта.

В июне 2020 года разработчики обсудили форк эффективно обновляемой нейронной сети (NNUE), представленный компьютерными сёги- программистами под названием Stockfish NNUE. В июле 2020 года в новостях о шахматах сообщалось, что Stockfish NNUE «открыл новые возможности в компьютерных шахматах, включив нейронную сеть в и без того невероятно мощный шахматный движок Stockfish». Затем было объявлено о слиянии NNUE с Stockfish, и стали доступны сборки для разработки.

2 сентября 2020 года была выпущена двенадцатая версия Stockfish, включающая вышеупомянутое усовершенствование нейронной сети. Согласно объявлению в блоге, эта новая версия «играет значительно сильнее, чем любая из ее предшественников», обычно выигрывая в десять раз больше игр, чем проигрывает в сравнении с одиннадцатой версией.

Client-side analysis

Securely running programs like Stockfish in your browser requires them to be ported to JavaScript and/or WebAssembly (WASM). Previously this would disproportionately slow down NNUE. The slowdown has been fixed thanks to a recent contribution, so we can now provide Stockfish 13 NNUE for modern browsers.

Note that NNUE is most likely crunching fewer nodes per second than Stockfish with classical evaluation, but it is stronger nonetheless.

The implementation uses WASM SIMD, which allows efficiently applying the same CPU instruction to multiple pieces of data, speeding up the evaluation of the neural network.

When we last reported on the status of NNUE in WASM , the prototype was letting Emscripten choose the instructions based on the x86-targeted code, and failing to achieve good results. Now the proper WASM SIMD instructions are picked by hand using compiler intrinsics.

Computing a dot product with WASM SIMD intrinsics

Using NNUE will require a download of about 10 MB (20 MB uncompressed) for the neural network file. We will refrain from using NNUE, if your browser communicates that your device is in save data mode.

Update 6th Mar: There is now a «Use NNUE» toggle in the engine settings, that can override save data mode.

Using in Chromium, Chrome (except on iOS), and Edge

Requires version 88. WASM SIMD should work out of the box, because Lichess is participating in an Origin Trial for chrome://flags/#enable-webassembly-simd.

Update 23th Feb: Some Android users have reported that updating to Chrome 89 Beta was required.

Firefox

Firefox users will need more patience. If you’re feeling adventurous, you can enable  javascript.options.wasm_simd in about:config. We have tested this with Firefox 85, but note that Firefox does not guarantee stability or security when tinkering with these flags.

Other browsers

Other browsers do not yet support WASM SIMD and will fall back to other Stockfish builds. We do not plan to update the fallbacks, so the gap in strength will grow as Stockfish progresses. This applies also to classical evaluation in newer Stockfish versions, since NNUE-specific search optimizations can be detrimental to classical evaluation strength. Here is the definitive list of available builds on Lichess, best to worst.

Update 23th Feb: Downgraded HCE build from Stockfish 12 to the optimal branching point for handcrafted evaluation (SF_classical). Later NNUE-specific search optimizations would lose Elo when not using NNUE.

Terms of use

Stockfish is free, and distributed under the GNU General Public License version 3
(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 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 the GPL.

For full details, read the copy of the GPL v3 found in the file named
Copying.txt.

Функции

Stockfish может использовать до 512 потоков ЦП в многопроцессорных системах. Максимальный размер его таблицы транспонирования — 32 ТБ. Stockfish реализует расширенный альфа-бета-поиск и использует битовые доски . По сравнению с другими движками, он отличается большой глубиной поиска, отчасти из-за более агрессивной обрезки и позднего сокращения перемещений . По состоянию на сентябрь 2021 года Stockfish 14 (4-х ниточный) достигает рейтинга Эло 3550.+22 ₋₂₂ на тесте CCRL 40/15.

Stockfish поддерживает Chess960 , что является одной из функций, унаследованных от Glaurung.

Поддержка Syzygy tablebase , ранее доступная в форке, поддерживаемом Рональдом де Маном, была интегрирована в Stockfish в 2014 году. В 2018 году была добавлена ​​поддержка Syzygy для 7 человек, вскоре после того, как она стала доступной.

Large Pages

Stockfish supports large pages on Linux and Windows. Large pages make
the hash access more efficient, improving the engine speed, especially
on large hash sizes. Typical increases are 5..10% in terms of nodes per
second, but speed increases up to 30% have been measured. The support is
automatic. Stockfish attempts to use large pages when available and
will fall back to regular memory allocation when this is not the case.

Support on Linux

Large page support on Linux is obtained by the Linux kernel
transparent huge pages functionality. Typically, transparent huge pages
are already enabled, and no configuration is needed.

Training Guide

Generating Training Data

Use the «no-nnue.nnue-gen-sfen-from-original-eval» binary. The given example is generation in its simplest form. There are more commands.

This will save a file named «generated_kifu.bin» in the same folder as the binary. Once generation is done, rename the file to something like «1billiondepth12.bin» to remember the depth and quantity of the positions and move it to a folder named «trainingdata» in the same directory as the binaries.

Generation Parameters

• Depth is the searched depth per move, or how far the engine looks forward. This value is an integer.
• Loop is the amount of positions generated. This value is also an integer

Generating Validation Data

The process is the same as the generation of training data, except for the fact that you need to set loop to 1 million, because you don’t need a lot of validation data. The depth should be the same as before or slightly higher than the depth of the training data. After generation rename the validation data file to val.bin and drop it in a folder named «validationdata» in the same directory to make it easier.

Training a Completely New Network

Use the «avx2.halfkp_256x2-32-32.nnue-learn.2020-07-11» binary. Create an empty folder named «evalsave» in the same directory as the binaries.

Nets get saved in the «evalsave» folder.

Training Parameters

• eta is the learning rate
• lambda is the amount of weight it puts to eval of learning data vs win/draw/loss results. 1 puts all weight on eval, lambda 0 puts all weight on WDL results.

Reinforcement Learning

If you would like to do some reinforcement learning on your original network, you must first generate training data using the learn binaries. Make sure that your previously trained network is in the eval folder. Use the commands specified above. Make sure is set to false so that the data generated is using the neural net’s eval by typing the command before typing the command. You should aim to generate less positions than the first run, around 1/10 of the number of positions generated in the first run. The depth should be higher as well. You should also do the same for validation data, with the depth being higher than the last run.

After you have generated the training data, you must move it into your training data folder and delete the older data so that the binary does not accidentally train on the same data again. Do the same for the validation data and name it to val-1.bin to make it less confusing. Make sure the evalsave folder is empty. Then, using the same binary, type in the training commands shown above. Do NOT set to true, it must be false or you will get a completely new network, instead of a network trained with reinforcement learning. You should also set eval_save_interval to a number that is lower than the amount of positions in your training data, perhaps also 1/10 of the original value. The validation file should be set to the new validation data, not the old data.

After training is finished, your new net should be located in the «final» folder under the «evalsave» directory. You should test this new network against the older network to see if there are any improvements.

Фиштест

С 2013 года Stockfish разрабатывается с использованием распределенной среды тестирования Fishtest , где добровольцы могут жертвовать процессорное время для тестирования улучшений программы.

Изменения в игровом коде принимаются или отклоняются на основании результатов прохождения десятков тысяч игр на платформе против более старой «эталонной» версии программы с использованием последовательного тестирования отношения вероятностей . Тесты на платформе проверяются с помощью теста хи-квадрат , и только если результаты статистически значимы, они считаются надежными и используются для корректировки программного кода.

По состоянию на июнь 2021 года фреймворк использовал в общей сложности более 5400 лет процессорного времени, чтобы сыграть более 3,1 миллиарда шахматных партий. После создания Fishtest всего за 12 месяцев Stockfish испытала взрывной рост на 120 баллов Эло , подняв его на вершину всех основных рейтинговых списков. В Stockfish 7 автор Fishtest Гэри Линскотт был добавлен в официальный список авторов в знак признания его вклада в силу Stockfish.

The UCI protocol and available options

Developers can see the default values for UCI options available in Stockfish by typing
in a terminal, but the majority of users will typically see them and
change them via a chess GUI. This is a list of available UCI options in Stockfish:

• The number of CPU threads used for searching a position. For best performance, set
  this equal to the number of CPU cores available.

• The size of the hash table in MB. It is recommended to set Hash after setting Threads.

• Clear the hash table.

• Let Stockfish ponder its next move while the opponent is thinking.

• Output the N best lines (principal variations, PVs) when searching.
  Leave at 1 for best performance.

• Toggle between the NNUE and classical evaluation functions. If set to «true»,
  the network parameters must be available to load from file (see also EvalFile),
  if they are not embedded in the binary.

• The name of the file of the NNUE evaluation parameters. Depending on the GUI the
  filename might have to include the full path to the folder/directory that contains the file.
  Other locations, such as the directory that contains the binary and the working directory,
  are also searched.

• An option handled by your GUI.

• An option handled by your GUI. If true, Stockfish will play Chess960.

• If enabled, show approximate WDL statistics as part of the engine output.
  These WDL numbers model expected game outcomes for a given evaluation and
  game ply for engine self-play at fishtest LTC conditions (60+0.6s per game).

• Enable weaker play aiming for an Elo rating as set by UCI_Elo. This option overrides Skill Level.

• If enabled by UCI_LimitStrength, aim for an engine strength of the given Elo.
  This Elo rating has been calibrated at a time control of 60s+0.6s and anchored to CCRL 40/4.

• Lower the Skill Level in order to make Stockfish play weaker (see also UCI_LimitStrength).
  Internally, MultiPV is enabled, and with a certain probability depending on the Skill Level a
  weaker move will be played.

• SyzygyPath

  Path to the folders/directories storing the Syzygy tablebase files. Multiple
  directories are to be separated by «;» on Windows and by «:» on Unix-based
  operating systems. Do not use spaces around the «;» or «:».

  Example:

  It is recommended to store .rtbw files on an SSD. There is no loss in storing
  the .rtbz files on a regular HD. It is recommended to verify all md5 checksums
  of the downloaded tablebase files () as corruption will
  lead to engine crashes.

• Minimum remaining search depth for which a position is probed. Set this option
  to a higher value to probe less aggressively if you experience too much slowdown
  (in terms of nps) due to tablebase probing.

• Disable to let fifty-move rule draws detected by Syzygy tablebase probes count
  as wins or losses. This is useful for ICCF correspondence games.

• Limit Syzygy tablebase probing to positions with at most this many pieces left
  (including kings and pawns).

• Assume a time delay of x ms due to network and GUI overheads. This is useful to
  avoid losses on time in those cases.

• Lower values will make Stockfish take less time in games, higher values will
  make it think longer.

• Tells the engine to use nodes searched instead of wall time to account for
  elapsed time. Useful for engine testing.

• Write all communication to and from the engine into a text file.

For developers the following non-standard commands might be of interest, mainly useful for debugging:

• Performs a standard benchmark using various options. The signature of a version (standard node
  count) is obtained using all defaults. is currently .

• Give information about the compiler and environment used for building a binary.

• Display the current position, with ascii art and fen.

• Return the evaluation of the current position.

• Exports the currently loaded network to a file.
  If the currently loaded network is the embedded network and the filename
  is not specified then the network is saved to the file matching the name
  of the embedded network, as defined in evaluate.h.
  If the currently loaded network is not the embedded network (some net set
  through the UCI setoption) then the filename parameter is required and the
  network is saved into that file.

Compiling Stockfish yourself from the sources

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
. In general it is recommended to run to see a list of make
targets with corresponding descriptions.

When not using the Makefile to compile (for instance, with Microsoft MSVC) you
need to manually set/unset some switches in the compiler command line; see
file types.h for a quick reference.

When reporting an issue or a bug, please tell us which Stockfish version
and which compiler you used to create your executable. This information
can be found by typing the following command in a console:

StockFish

Описание:Hidden text. Click here

Hidden text. Click here

ТОП-50 движков

Платформы

Релизные и разрабатываемые версии доступны как исходный код C ++ и как предварительно скомпилированные версии для Microsoft Windows , macOS , 32-разрядной / 64-разрядной версии Linux и Android .

Stockfish был очень популярным движком на различных платформах. На настольных компьютерах это шахматный движок по умолчанию, связанный с интерфейсными программами интернет-шахматного клуба BlitzIn и Dasher. На мобильных устройствах он поставляется в комплекте с приложением Stockfish, SmallFish и Droidfish. Другие графические пользовательские интерфейсы, совместимые со Stockfish, включают Fritz , Arena, Stockfish для Mac и PyChess . По состоянию на март 2014 года Stockfish — это шахматный движок, используемый Lichess , популярным шахматным онлайн-сайтом.

Stockfish можно скомпилировать в WebAssembly или JavaScript , что позволит запускать его в браузере. И Chess.com, и Lichess предоставляют Stockfish в этой форме в дополнение к серверной программе.

What to expect from the Syzygy tablebases?

If the engine is searching a position that is not in the tablebases (e.g.
a position with 8 pieces), it will access the tablebases during the search.
If the engine reports a very large score (typically 153.xx), this means
it has found a winning line into a tablebase position.

If the engine is given a position to search that is in the tablebases, it
will use the tablebases at the beginning of the search to preselect all
good moves, i.e. all moves that preserve the win or preserve the draw while
taking into account the 50-move rule.
It will then perform a search only on those moves. The engine will not move
immediately, unless there is only a single good move. The engine likely
will not report a mate score, even if the position is known to be won.

It is therefore clear that this behaviour is not identical to what one might
be used to with Nalimov tablebases. There are technical reasons for this
difference, the main technical reason being that Nalimov tablebases use the
DTM metric (distance-to-mate), while the Syzygy tablebases use a variation of the
DTZ metric (distance-to-zero, zero meaning any move that resets the 50-move
counter). This special metric is one of the reasons that the Syzygy tablebases are
more compact than Nalimov tablebases, while still storing all information
needed for optimal play and in addition being able to take into account
the 50-move rule.

A note on classical evaluation versus NNUE evaluation

Both approaches assign a value to a position that is used in alpha-beta (PVS) search
to find the best move. The classical evaluation computes this value as a function
of various chess concepts, handcrafted by experts, tested and tuned using fishtest.
The NNUE evaluation computes this value with a neural network based on basic
inputs (e.g. piece positions only). The network is optimized and trained
on the evaluations of millions of positions at moderate search depth.

The NNUE evaluation was first introduced in shogi, and ported to Stockfish afterward.
It can be evaluated efficiently on CPUs, and exploits the fact that only parts
of the neural network need to be updated after a typical chess move.
The nodchip repository provides additional
tools to train and develop the NNUE networks. On CPUs supporting modern vector instructions
(avx2 and similar), the NNUE evaluation results in much stronger playing strength, even
if the nodes per second computed by the engine is somewhat lower (roughly 80% of nps
is typical).

Notes:

1. the NNUE evaluation depends on the Stockfish binary and the network parameter
   file (see the EvalFile UCI option). Not every parameter file is compatible with a given
   Stockfish binary, but the default value of the EvalFile UCI option is the name of a network
   that is guaranteed to be compatible with that binary.

2. to use the NNUE evaluation, the additional data file with neural network parameters
   needs to be available. Normally, this file is already embedded in the binary or it
   can be downloaded. The filename for the default (recommended) net can be found as the default
   value of the UCI option, with the format
   (for instance, ). This file can be downloaded from

replacing as needed.