3 [![Build Status](https://travis-ci.org/official-stockfish/Stockfish.svg?branch=master)](https://travis-ci.org/official-stockfish/Stockfish)
4 [![Build Status](https://ci.appveyor.com/api/projects/status/github/official-stockfish/Stockfish?branch=master&svg=true)](https://ci.appveyor.com/project/mcostalba/stockfish/branch/master)
6 [Stockfish](https://stockfishchess.org) is a free, powerful UCI chess engine
7 derived from Glaurung 2.1. Stockfish is not a complete chess program and requires a
8 UCI-compatible graphical user interface (GUI) (e.g. XBoard with PolyGlot, Scid,
9 Cute Chess, eboard, Arena, Sigma Chess, Shredder, Chess Partner or Fritz) in order
10 to be used comfortably. Read the documentation for your GUI of choice for information
11 about how to use Stockfish with it.
13 The Stockfish engine features two evaluation functions for chess, the classical
14 evaluation based on handcrafted terms, and the NNUE evaluation based on efficiently
15 updatable neural networks. The classical evaluation runs efficiently on almost all
16 CPU architectures, while the NNUE evaluation benefits from the vector
17 intrinsics available on most CPUs (sse2, avx2, neon, or similar).
22 This distribution of Stockfish consists of the following files:
24 * Readme.md, the file you are currently reading.
26 * Copying.txt, a text file containing the GNU General Public License version 3.
28 * AUTHORS, a text file with the list of authors for the project
30 * src, a subdirectory containing the full source code, including a Makefile
31 that can be used to compile Stockfish on Unix-like systems.
33 * a file with the .nnue extension, storing the neural network for the NNUE
34 evaluation. Binary distributions will have this file embedded.
36 ## The UCI protocol and available options
38 The Universal Chess Interface (UCI) is a standard protocol used to communicate with a chess engine,
39 and is the recommended way to do so for typical graphical user interfaces (GUI) or chess tools.
41 Stockfish implements most commands as described in [the UCI protocol](https://www.shredderchess.com/download/div/uci.zip)
43 For users, the following UCI options, which can typically be set via a GUI, are available in Stockfish:
46 The number of CPU threads used for searching a position. For best performance, set
47 this equal to the number of CPU cores available.
50 The size of the hash table in MB. It is recommended to set Hash after setting Threads.
56 Let Stockfish ponder its next move while the opponent is thinking.
59 Output the N best lines (principal variations, PVs) when searching.
60 Leave at 1 for best performance.
63 Toggle between the NNUE and classical evaluation functions. If set to "true",
64 the network parameters must be available to load from file (see also EvalFile),
65 if they are not embedded in the binary.
68 The name of the file of the NNUE evaluation parameters. Depending on the GUI the
69 filename might have to include the full path to the folder/directory that contains the file.
70 Other locations, such as the directory that contains the binary and the working directory,
73 * #### UCI_AnalyseMode
74 An option handled by your GUI.
77 An option handled by your GUI. If true, Stockfish will play Chess960.
80 If enabled, show approximate WDL statistics as part of the engine output.
81 These WDL numbers model expected game outcomes for a given evaluation and
82 game ply for engine self-play at fishtest LTC conditions (60+0.6s per game).
84 * #### UCI_LimitStrength
85 Enable weaker play aiming for an Elo rating as set by UCI_Elo. This option overrides Skill Level.
88 If enabled by UCI_LimitStrength, aim for an engine strength of the given Elo.
89 This Elo rating has been calibrated at a time control of 60s+0.6s and anchored to CCRL 40/4.
92 Lower the Skill Level in order to make Stockfish play weaker (see also UCI_LimitStrength).
93 Internally, MultiPV is enabled, and with a certain probability depending on the Skill Level a
94 weaker move will be played.
97 Path to the folders/directories storing the Syzygy tablebase files. Multiple
98 directories are to be separated by ";" on Windows and by ":" on Unix-based
99 operating systems. Do not use spaces around the ";" or ":".
101 Example: `C:\tablebases\wdl345;C:\tablebases\wdl6;D:\tablebases\dtz345;D:\tablebases\dtz6`
103 It is recommended to store .rtbw files on an SSD. There is no loss in storing
104 the .rtbz files on a regular HD. It is recommended to verify all md5 checksums
105 of the downloaded tablebase files (`md5sum -c checksum.md5`) as corruption will
106 lead to engine crashes.
108 * #### SyzygyProbeDepth
109 Minimum remaining search depth for which a position is probed. Set this option
110 to a higher value to probe less aggressively if you experience too much slowdown
111 (in terms of nps) due to tablebase probing.
113 * #### Syzygy50MoveRule
114 Disable to let fifty-move rule draws detected by Syzygy tablebase probes count
115 as wins or losses. This is useful for ICCF correspondence games.
117 * #### SyzygyProbeLimit
118 Limit Syzygy tablebase probing to positions with at most this many pieces left
119 (including kings and pawns).
122 A positive value for contempt favors middle game positions and avoids draws,
123 effective for the classical evaluation only.
125 * #### Analysis Contempt
126 By default, contempt is set to prefer the side to move. Set this option to "White"
127 or "Black" to analyse with contempt for that side, or "Off" to disable contempt.
130 Assume a time delay of x ms due to network and GUI overheads. This is useful to
131 avoid losses on time in those cases.
134 Lower values will make Stockfish take less time in games, higher values will
135 make it think longer.
138 Tells the engine to use nodes searched instead of wall time to account for
139 elapsed time. Useful for engine testing.
141 * #### Debug Log File
142 Write all communication to and from the engine into a text file.
144 For developers the following non-standard commands might be of interest, mainly useful for debugging:
146 * #### bench ttSize threads limit fenFile limitType evalType
147 Performs a standard benchmark using various options. The signature or standard node
148 count is obtained using all defaults. `bench` is currently `bench 16 1 13 default depth mixed`.
151 Give information about the compiler and environment used for building a binary.
154 Display the current position, with ascii art and fen.
157 Return the evaluation of the current position.
159 * #### export_net [filename]
160 Exports the currently loaded network to a file.
161 If the currently loaded network is the embedded network and the filename
162 is not specified then the network is saved to the file matching the name
163 of the embedded network, as defined in evaluate.h.
164 If the currently loaded network is not the embedded network (some net set
165 through the UCI setoption) then the filename parameter is required and the
166 network is saved into that file.
169 Flips the side to move.
172 ## A note on classical evaluation versus NNUE evaluation
174 Both approaches assign a value to a position that is used in alpha-beta (PVS) search
175 to find the best move. The classical evaluation computes this value as a function
176 of various chess concepts, handcrafted by experts, tested and tuned using fishtest.
177 The NNUE evaluation computes this value with a neural network based on basic
178 inputs (e.g. piece positions only). The network is optimized and trained
179 on the evaluations of millions of positions at moderate search depth.
181 The NNUE evaluation was first introduced in shogi, and ported to Stockfish afterward.
182 It can be evaluated efficiently on CPUs, and exploits the fact that only parts
183 of the neural network need to be updated after a typical chess move.
184 [The nodchip repository](https://github.com/nodchip/Stockfish) provides additional
185 tools to train and develop the NNUE networks. On CPUs supporting modern vector instructions
186 (avx2 and similar), the NNUE evaluation results in much stronger playing strength, even
187 if the nodes per second computed by the engine is somewhat lower (roughly 80% of nps
192 1) the NNUE evaluation depends on the Stockfish binary and the network parameter
193 file (see the EvalFile UCI option). Not every parameter file is compatible with a given
194 Stockfish binary, but the default value of the EvalFile UCI option is the name of a network
195 that is guaranteed to be compatible with that binary.
197 2) to use the NNUE evaluation, the additional data file with neural network parameters
198 needs to be available. Normally, this file is already embedded in the binary or it
199 can be downloaded. The filename for the default (recommended) net can be found as the default
200 value of the `EvalFile` UCI option, with the format `nn-[SHA256 first 12 digits].nnue`
201 (for instance, `nn-c157e0a5755b.nnue`). This file can be downloaded from
203 https://tests.stockfishchess.org/api/nn/[filename]
205 replacing `[filename]` as needed.
207 ## What to expect from the Syzygy tablebases?
209 If the engine is searching a position that is not in the tablebases (e.g.
210 a position with 8 pieces), it will access the tablebases during the search.
211 If the engine reports a very large score (typically 153.xx), this means
212 it has found a winning line into a tablebase position.
214 If the engine is given a position to search that is in the tablebases, it
215 will use the tablebases at the beginning of the search to preselect all
216 good moves, i.e. all moves that preserve the win or preserve the draw while
217 taking into account the 50-move rule.
218 It will then perform a search only on those moves. **The engine will not move
219 immediately**, unless there is only a single good move. **The engine likely
220 will not report a mate score, even if the position is known to be won.**
222 It is therefore clear that this behaviour is not identical to what one might
223 be used to with Nalimov tablebases. There are technical reasons for this
224 difference, the main technical reason being that Nalimov tablebases use the
225 DTM metric (distance-to-mate), while the Syzygy tablebases use a variation of the
226 DTZ metric (distance-to-zero, zero meaning any move that resets the 50-move
227 counter). This special metric is one of the reasons that the Syzygy tablebases are
228 more compact than Nalimov tablebases, while still storing all information
229 needed for optimal play and in addition being able to take into account
234 Stockfish supports large pages on Linux and Windows. Large pages make
235 the hash access more efficient, improving the engine speed, especially
236 on large hash sizes. Typical increases are 5..10% in terms of nodes per
237 second, but speed increases up to 30% have been measured. The support is
238 automatic. Stockfish attempts to use large pages when available and
239 will fall back to regular memory allocation when this is not the case.
243 Large page support on Linux is obtained by the Linux kernel
244 transparent huge pages functionality. Typically, transparent huge pages
245 are already enabled, and no configuration is needed.
247 ### Support on Windows
249 The use of large pages requires "Lock Pages in Memory" privilege. See
250 [Enable the Lock Pages in Memory Option (Windows)](https://docs.microsoft.com/en-us/sql/database-engine/configure-windows/enable-the-lock-pages-in-memory-option-windows)
251 on how to enable this privilege, then run [RAMMap](https://docs.microsoft.com/en-us/sysinternals/downloads/rammap)
252 to double-check that large pages are used. We suggest that you reboot
253 your computer after you have enabled large pages, because long Windows
254 sessions suffer from memory fragmentation, which may prevent Stockfish
255 from getting large pages: a fresh session is better in this regard.
257 ## Compiling Stockfish yourself from the sources
259 Stockfish has support for 32 or 64-bit CPUs, certain hardware
260 instructions, big-endian machines such as Power PC, and other platforms.
262 On Unix-like systems, it should be easy to compile Stockfish
263 directly from the source code with the included Makefile in the folder
264 `src`. In general it is recommended to run `make help` to see a list of make
265 targets with corresponding descriptions.
271 make build ARCH=x86-64-modern
274 When not using the Makefile to compile (for instance, with Microsoft MSVC) you
275 need to manually set/unset some switches in the compiler command line; see
276 file *types.h* for a quick reference.
278 When reporting an issue or a bug, please tell us which Stockfish version
279 and which compiler you used to create your executable. This information
280 can be found by typing the following command in a console:
286 ## Understanding the code base and participating in the project
288 Stockfish's improvement over the last decade has been a great community
289 effort. There are a few ways to help contribute to its growth.
291 ### Donating hardware
293 Improving Stockfish requires a massive amount of testing. You can donate
294 your hardware resources by installing the [Fishtest Worker](https://github.com/glinscott/fishtest/wiki/Running-the-worker:-overview)
295 and view the current tests on [Fishtest](https://tests.stockfishchess.org/tests).
297 ### Improving the code
299 If you want to help improve the code, there are several valuable resources:
301 * [In this wiki,](https://www.chessprogramming.org) many techniques used in
302 Stockfish are explained with a lot of background information.
304 * [The section on Stockfish](https://www.chessprogramming.org/Stockfish)
305 describes many features and techniques used by Stockfish. However, it is
306 generic rather than being focused on Stockfish's precise implementation.
307 Nevertheless, a helpful resource.
309 * The latest source can always be found on [GitHub](https://github.com/official-stockfish/Stockfish).
310 Discussions about Stockfish take place these days mainly in the [FishCooking](https://groups.google.com/forum/#!forum/fishcooking)
311 group and on the [Stockfish Discord channel](https://discord.gg/nv8gDtt).
312 The engine testing is done on [Fishtest](https://tests.stockfishchess.org/tests).
313 If you want to help improve Stockfish, please read this [guideline](https://github.com/glinscott/fishtest/wiki/Creating-my-first-test)
314 first, where the basics of Stockfish development are explained.
319 Stockfish is free, and distributed under the **GNU General Public License version 3**
320 (GPL v3). Essentially, this means you are free to do almost exactly
321 what you want with the program, including distributing it among your
322 friends, making it available for download from your website, selling
323 it (either by itself or as part of some bigger software package), or
324 using it as the starting point for a software project of your own.
326 The only real limitation is that whenever you distribute Stockfish in
327 some way, you MUST always include the full source code, or a pointer
328 to where the source code can be found, to generate the exact binary
329 you are distributing. If you make any changes to the source code,
330 these changes must also be made available under the GPL.
332 For full details, read the copy of the GPL v3 found in the file named