Difference between revisions of "Seer"
GerdIsenberg (talk | contribs) (Created page with "'''Home * Engines * Seer''' FILE:Old seer - olympia.jpg|border|right|thumb|240px| Greek Seer <ref>[https://commons.wikimedia.org/wiki/File:GR_08-04-23_Oly...") |
GerdIsenberg (talk | contribs) |
||
(13 intermediate revisions by the same user not shown) | |||
Line 1: | Line 1: | ||
'''[[Main Page|Home]] * [[Engines]] * Seer''' | '''[[Main Page|Home]] * [[Engines]] * Seer''' | ||
− | [[FILE:Old seer - olympia.jpg|border|right|thumb|240px| Greek Seer <ref>[https://commons.wikimedia.org/wiki/File:GR_08-04-23_Olympia_Museum_Zeustempel_Ostgiebel2.JPG | + | [[FILE:Old seer - olympia.jpg|border|right|thumb|240px| Greek Seer <ref>[https://commons.wikimedia.org/wiki/File:GR_08-04-23_Olympia_Museum_Zeustempel_Ostgiebel2.JPG Detail] of [https://en.wikipedia.org/wiki/Temple_of_Zeus,_Olympia Temple of Zeus, Olympia], author [https://commons.wikimedia.org/wiki/Category:Fotos_by_Monika_Angela_Arnold,_Berlin Monika Angela Arnold, Berlin], [https://en.wikipedia.org/wiki/Wikimedia_Commons Wikimedia Commons]</ref> ]] |
'''Seer''',<br/> | '''Seer''',<br/> | ||
an [[UCI]] compliant [[:Category:Open Source|open source chess engine]] by [[Connor McMonigle]], written in [[Cpp|C++]], | an [[UCI]] compliant [[:Category:Open Source|open source chess engine]] by [[Connor McMonigle]], written in [[Cpp|C++]], | ||
licensed under the [[Free Software Foundation#GPL|GNU GPL]] and first released in October 2020. | licensed under the [[Free Software Foundation#GPL|GNU GPL]] and first released in October 2020. | ||
− | Seer uses a custom [[NNUE]] implementation with training code written in [https://en.wikipedia.org/wiki/PyTorch PyTorch] and inference code relying on [https://en.wikipedia.org/wiki/OpenMP OpenMP] [[SIMD and SWAR Techniques|SIMD]] for auto vectorization <ref>[https://github.com/connormcmonigle/seer-nnue GitHub - connormcmonigle/seer-nnue: UCI chess engine using NNUE for position evaluation]</ref>. | + | Seer uses a custom [[NNUE]] implementation based on [[Float|32-bit float]] weights with training code written in [https://en.wikipedia.org/wiki/PyTorch PyTorch] and inference code relying on [https://en.wikipedia.org/wiki/OpenMP OpenMP] [[SIMD and SWAR Techniques|SIMD]] for auto vectorization <ref>[https://github.com/connormcmonigle/seer-nnue GitHub - connormcmonigle/seer-nnue: UCI chess engine using NNUE for position evaluation]</ref>. |
+ | Due to [[BMI2#PEXTBitboards|PEXT Bitboards]], Seer requires [[BMI2]] for a reasonable performance, as well as either [[SSE]], [[AVX]], [[AVX2]] or [[AVX-512]] to calculate the NNUE. More recently, Seer supports [[ARM NEON]] via '''sse2neon''' <ref>[https://github.com/DLTcollab/sse2neon GitHub - DLTcollab/sse2neon: A translator from Intel SSE intrinsics to Arm/Aarch64 NEON implementation]</ref> to run on many [[Android]] devices <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=77187&start=32 Re: Seer 2.0.0] by Archimedes, [[CCC]], August 06, 2021</ref>. | ||
=Features= | =Features= | ||
Line 23: | Line 24: | ||
* [[Move Ordering]] | * [[Move Ordering]] | ||
** [[Static Exchange Evaluation|SEE]] for [[Captures]] | ** [[Static Exchange Evaluation|SEE]] for [[Captures]] | ||
+ | ** [[Killer Heuristic]] | ||
** [[History Heuristic]] | ** [[History Heuristic]] | ||
** [[History Heuristic#CMHist|Counter Moves History]] | ** [[History Heuristic#CMHist|Counter Moves History]] | ||
Line 39: | Line 41: | ||
* [[Oracle]] | * [[Oracle]] | ||
* [[Prophet]] | * [[Prophet]] | ||
+ | * [[Minic#Minic 3|Minic 3]] | ||
=Forum Posts= | =Forum Posts= | ||
+ | ==2020== | ||
* [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72613&start=427 Re: New engine releases 2020] by [[Connor McMonigle]], [[CCC]], October 18, 2020 | * [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72613&start=427 Re: New engine releases 2020] by [[Connor McMonigle]], [[CCC]], October 18, 2020 | ||
+ | * [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75433 Seer] by [[Gerd Isenberg]], [[CCC]], October 18, 2020 | ||
+ | : [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75433&start=58 Re: Seer] by [[Connor McMonigle]], [[CCC]], November 02, 2020 » Seer 1.1 | ||
+ | : [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75433&start=109 Re: Seer] by [[Connor McMonigle]], [[CCC]], November 18, 2020 » Seer 1.2 | ||
+ | * [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=72613&start=469 Re: New engine releases 2020] by [[Connor McMonigle]], [[CCC]], November 02, 2020 | ||
+ | * [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75665&start=9 Re: Minic version 3] by [[Connor McMonigle]], [[CCC]], November 03, 2020 » [[Minic#Minic 3|Minic 3]] <ref>[[Gao Huang]], [[Zhuang Liu]], [[Laurens van der Maaten]], [[Kilian Q. Weinberger]] ('''2016'''). ''Densely Connected Convolutional Networks''. [https://arxiv.org/abs/1608.06993 arXiv:1608.06993]</ref> | ||
+ | * [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75890&start=6 Re: Speculations about NNUE development (was New engine releases 2020)] by [[Connor McMonigle]], [[CCC]], November 12, 2020 » [[Dragon by Komodo Chess]], [[Halogen]] | ||
+ | : [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=75890&start=9 Re: Speculations about NNUE development (was New engine releases 2020)] by [[Connor McMonigle]], [[CCC]], November 12, 2020 | ||
+ | ==2021 ...== | ||
+ | * [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=77187 Seer 2.0.0] by [[Connor McMonigle]], [[CCC]], April 27, 2021 | ||
+ | : [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=77187&start=32 Re: Seer 2.0.0] by Archimedes, [[CCC]], August 06, 2021 » Seer 2.2.0 for [[Android]] | ||
+ | : [https://www.talkchess.com/forum3/viewtopic.php?f=2&t=77187&start=60 Re: Seer 2.0.0] by [[Connor McMonigle]], [[CCC]], April 25, 2022 » Seer 2.5.0 | ||
+ | * [http://www.talkchess.com/forum3/viewtopic.php?f=2&t=77503&start=55 Re: will Tcec allow Stockfish with a Leela net to play?] by [[Connor McMonigle]], [[CCC]], June 17, 2021 » [[NNUE]] | ||
=External Links= | =External Links= | ||
==Chess Engine== | ==Chess Engine== | ||
* [https://github.com/connormcmonigle/seer-nnue GitHub - connormcmonigle/seer-nnue: UCI chess engine using NNUE for position evaluation] | * [https://github.com/connormcmonigle/seer-nnue GitHub - connormcmonigle/seer-nnue: UCI chess engine using NNUE for position evaluation] | ||
+ | * [https://github.com/connormcmonigle/seer-nnue/releases/tag/v2.5.0 Release seer-v2.5.0 · connormcmonigle/seer-nnue · GitHub] | ||
==Misc== | ==Misc== | ||
* [https://en.wiktionary.org/wiki/seer seer - Wiktionary] | * [https://en.wiktionary.org/wiki/seer seer - Wiktionary] | ||
* [https://en.wikipedia.org/wiki/Seer Seer from Wikipedia] | * [https://en.wikipedia.org/wiki/Seer Seer from Wikipedia] | ||
− | * [https://en.wikipedia.org/wiki/Hypnotic_Brass_Ensemble Hypnotic Brass Ensemble] - Seer, | + | * [https://en.wikipedia.org/wiki/Hypnotic_Brass_Ensemble Hypnotic Brass Ensemble] - Seer, [https://www.birdistheworm.com/recommended-hypnotic-brass-ensemble-sound-rhythm-form/ Sound Rhythm & Form] (2016), [https://en.wikipedia.org/wiki/YouTube YouTube] Video |
{{#evu:https://www.youtube.com/watch?v=KModymmEanc|alignment=left|valignment=top}} | {{#evu:https://www.youtube.com/watch?v=KModymmEanc|alignment=left|valignment=top}} | ||
Line 62: | Line 79: | ||
[[Category:Windows]] | [[Category:Windows]] | ||
[[Category:Linux]] | [[Category:Linux]] | ||
+ | [[Category:Android]] | ||
[[Category:NNUE]] | [[Category:NNUE]] | ||
[[Category:Metaphysics]] | [[Category:Metaphysics]] | ||
[[Category:Music]] | [[Category:Music]] |
Latest revision as of 20:47, 25 April 2022
Seer,
an UCI compliant open source chess engine by Connor McMonigle, written in C++,
licensed under the GNU GPL and first released in October 2020.
Seer uses a custom NNUE implementation based on 32-bit float weights with training code written in PyTorch and inference code relying on OpenMP SIMD for auto vectorization [2].
Due to PEXT Bitboards, Seer requires BMI2 for a reasonable performance, as well as either SSE, AVX, AVX2 or AVX-512 to calculate the NNUE. More recently, Seer supports ARM NEON via sse2neon [3] to run on many Android devices [4].
Features
Board Representation
Search
- Iterative Deepening
- Aspiration Windows
- Lazy SMP
- Principal Variation Search
- Transposition Table
- Move Ordering
- Selectivity
Evaluation
See also
Forum Posts
2020
- Re: New engine releases 2020 by Connor McMonigle, CCC, October 18, 2020
- Seer by Gerd Isenberg, CCC, October 18, 2020
- Re: Seer by Connor McMonigle, CCC, November 02, 2020 » Seer 1.1
- Re: Seer by Connor McMonigle, CCC, November 18, 2020 » Seer 1.2
- Re: New engine releases 2020 by Connor McMonigle, CCC, November 02, 2020
- Re: Minic version 3 by Connor McMonigle, CCC, November 03, 2020 » Minic 3 [6]
- Re: Speculations about NNUE development (was New engine releases 2020) by Connor McMonigle, CCC, November 12, 2020 » Dragon by Komodo Chess, Halogen
- Re: Speculations about NNUE development (was New engine releases 2020) by Connor McMonigle, CCC, November 12, 2020
2021 ...
- Seer 2.0.0 by Connor McMonigle, CCC, April 27, 2021
- Re: Seer 2.0.0 by Archimedes, CCC, August 06, 2021 » Seer 2.2.0 for Android
- Re: Seer 2.0.0 by Connor McMonigle, CCC, April 25, 2022 » Seer 2.5.0
- Re: will Tcec allow Stockfish with a Leela net to play? by Connor McMonigle, CCC, June 17, 2021 » NNUE
External Links
Chess Engine
- GitHub - connormcmonigle/seer-nnue: UCI chess engine using NNUE for position evaluation
- Release seer-v2.5.0 · connormcmonigle/seer-nnue · GitHub
Misc
- seer - Wiktionary
- Seer from Wikipedia
- Hypnotic Brass Ensemble - Seer, Sound Rhythm & Form (2016), YouTube Video
References
- ↑ Detail of Temple of Zeus, Olympia, author Monika Angela Arnold, Berlin, Wikimedia Commons
- ↑ GitHub - connormcmonigle/seer-nnue: UCI chess engine using NNUE for position evaluation
- ↑ GitHub - DLTcollab/sse2neon: A translator from Intel SSE intrinsics to Arm/Aarch64 NEON implementation
- ↑ Re: Seer 2.0.0 by Archimedes, CCC, August 06, 2021
- ↑ seer-nnue/README.md at master · connormcmonigle/seer-nnue · GitHub
- ↑ Gao Huang, Zhuang Liu, Laurens van der Maaten, Kilian Q. Weinberger (2016). Densely Connected Convolutional Networks. arXiv:1608.06993