New engine releases 2019
Moderators: hgm, Rebel, chrisw
-
- Posts: 182
- Joined: Sun Jun 12, 2016 5:44 pm
- Location: London
- Full name: Vincent
Re: New engine releases 2019
I'm just going to slide Topple v0.3.6 in here: https://github.com/konsolas/ToppleChess ... tag/v0.3.6
-
- Posts: 288
- Joined: Sat Jun 30, 2018 10:58 pm
- Location: Ukraine
- Full name: Volodymyr Shcherbyna
Re: New engine releases 2019
Hello,
Igel 1.3 is released at https://github.com/vshcherbyna/igel/releases/tag/1.3 . What's new:
- New eval terms for pair of rooks, knights, bishops, connected pawns
Approx. strength compared to Igel 1.2 is around 15 elo on single thread. I was also doing some tests to see performance improvements for lazy smp in Igel and it seems to be around 100 elo when using 11 threads against 1 thread:
threads=1 (for 1.2)
threads=11 (for 1.3)
hash=6GB
tc all/60+0.6
Score of Igel 1.2 64 POPCNT vs Igel 1.3 64 POPCNT: 38 - 138 - 97 [0.317] 273
Elo difference: -133.47 +/- 34.12
Igel 1.3 is released at https://github.com/vshcherbyna/igel/releases/tag/1.3 . What's new:
- New eval terms for pair of rooks, knights, bishops, connected pawns
Approx. strength compared to Igel 1.2 is around 15 elo on single thread. I was also doing some tests to see performance improvements for lazy smp in Igel and it seems to be around 100 elo when using 11 threads against 1 thread:
threads=1 (for 1.2)
threads=11 (for 1.3)
hash=6GB
tc all/60+0.6
Score of Igel 1.2 64 POPCNT vs Igel 1.3 64 POPCNT: 38 - 138 - 97 [0.317] 273
Elo difference: -133.47 +/- 34.12
-
- Posts: 1362
- Joined: Sat Jul 21, 2018 7:43 am
- Location: Szentendre, Hungary
- Full name: Gabor Szots
Re: New engine releases 2019
Gabor Szots
CCRL testing group
CCRL testing group
-
- Posts: 4605
- Joined: Wed Oct 01, 2008 6:33 am
- Location: Regensburg, Germany
- Full name: Guenther Simon
Re: New engine releases 2019
In the thread below for the upcoming GSEI tournament there are
several new releases (+betas) hidden :)
viewtopic.php?f=6&t=70084
several new releases (+betas) hidden :)
viewtopic.php?f=6&t=70084
-
- Posts: 68
- Joined: Fri Nov 16, 2018 10:23 am
- Full name: Laurens Winkelhagen
Re: New engine releases 2019
I'm happy to announce FrankWalter 2.3.5 - https://github.com/ljgw/frankwalter/releases/tag/v2.3.5
main new features:
* much improved LMR
* frankwalter.exe -> an executable wrapper generated by launch4j which should find and use the most suitable java runtime to run FrankWalter (can be used when people have problems setting up the FrankWalter java engine in their favorite GUI)
* longer PV shown in some specific situations
bugfixes:
* fixed a Transposition Table bug
* fixed inconsistent state after doMove / undoMove (obviously not a big inconsistency, but it has had its effects)
Results after 370 40/3 games on my machine:
main new features:
* much improved LMR
* frankwalter.exe -> an executable wrapper generated by launch4j which should find and use the most suitable java runtime to run FrankWalter (can be used when people have problems setting up the FrankWalter java engine in their favorite GUI)
* longer PV shown in some specific situations
bugfixes:
* fixed a Transposition Table bug
* fixed inconsistent state after doMove / undoMove (obviously not a big inconsistency, but it has had its effects)
Results after 370 40/3 games on my machine:
Code: Select all
Rank Name Elo +/- Games Score Draws
1 fw235 58 29 370 58.2% 33.8%
2 fwsnap 47 29 370 56.8% 33.5%
3 olithink 14 31 370 52.0% 25.7%
4 fw234 -123 30 370 33.0% 32.4%
Author of JanWillem (C, WB, inactive) and FrankWalter (Java, WB, https://github.com/ljgw/frankwalter)
-
- Posts: 1871
- Joined: Sat Nov 25, 2017 2:28 pm
- Location: France
Re: New engine releases 2019
Wow that is huge versus 2.3.4!ljgw wrote: ↑Sun Mar 10, 2019 11:59 am I'm happy to announce FrankWalter 2.3.5 - https://github.com/ljgw/frankwalter/releases/tag/v2.3.5
main new features:
* much improved LMR
* frankwalter.exe -> an executable wrapper generated by launch4j which should find and use the most suitable java runtime to run FrankWalter (can be used when people have problems setting up the FrankWalter java engine in their favorite GUI)
* longer PV shown in some specific situations
bugfixes:
* fixed a Transposition Table bug
* fixed inconsistent state after doMove / undoMove (obviously not a big inconsistency, but it has had its effects)
Results after 370 40/3 games on my machine:Code: Select all
Rank Name Elo +/- Games Score Draws 1 fw235 58 29 370 58.2% 33.8% 2 fwsnap 47 29 370 56.8% 33.5% 3 olithink 14 31 370 52.0% 25.7% 4 fw234 -123 30 370 33.0% 32.4%
-
- Posts: 68
- Joined: Fri Nov 16, 2018 10:23 am
- Full name: Laurens Winkelhagen
Re: New engine releases 2019
Well, I'm a bit skeptical; I should really invest in setting up some more opponents locally and run gauntlets:-)
but that there is an improvement is undeniable^^
but that there is an improvement is undeniable^^
Author of JanWillem (C, WB, inactive) and FrankWalter (Java, WB, https://github.com/ljgw/frankwalter)
-
- Posts: 4889
- Joined: Thu Mar 09, 2006 6:34 am
- Location: Pen Argyl, Pennsylvania
Re: New engine releases 2019
dev-McCain X2a (source only you must compile your own exe's) has been updated with both dev SF and McCain updates.
https://github.com/MichaelB7/Stockfish/tree/McCain
With the McCain source, you build both current devSF or current dev-McCain - which version you build is whether or not comment or uncomment line 34
### Uncomment the line below for McCain, comment out for Stockfish
VERSION=maverick
If you build current-dev-SF, it will also have all the bells and whistles for 3 opening books, play by Elo etc that McCain has
It played pretty well in a short match of 400 games at STC - ymmv naturally. In my opinion, this more is likely an aberration than any real indicator of strength. When it looks too good, it is usually not true. I have not introduced anything of significance lately that would justify this result. I did check the pgn field to make sure there were no time flags ( usually indicates a crash by one of the engines).
Of particular note, for those with macOS (N/A for other OS), it has this thread stack size bug update by Marco Costalba:
]https://github.com/official-stockfish/ ... /pull/2035
As an fyi, I'm working on a standalone mate finder type program (to be named "1st-Mate") which will combine both McCain and Matefinder by Joe Ellis - the source has not yet been released publicly, but when it is released it will never be as fast as Joe Ellis' matefinder in finding mates, but more of a hybrid engine between McCain and Matefinder.
Also, I have updated Joe Ellis' matefinder code with the latest dev-SF code to the best of my abilities ( again you must compile your own exe's). This might be of some interest since Joe's matefinder code has not been updated on GitHub since late December.
https://github.com/MichaelB7/Stockfish/tree/mate_finder
https://github.com/MichaelB7/Stockfish/tree/McCain
With the McCain source, you build both current devSF or current dev-McCain - which version you build is whether or not comment or uncomment line 34
### Uncomment the line below for McCain, comment out for Stockfish
VERSION=maverick
If you build current-dev-SF, it will also have all the bells and whistles for 3 opening books, play by Elo etc that McCain has
It played pretty well in a short match of 400 games at STC - ymmv naturally. In my opinion, this more is likely an aberration than any real indicator of strength. When it looks too good, it is usually not true. I have not introduced anything of significance lately that would justify this result. I did check the pgn field to make sure there were no time flags ( usually indicates a crash by one of the engines).
Code: Select all
400 rounds and 400 games completed...
Time control: 15 seconds + .500 seconds
Target completion: 03/11/19 : 00:04:29
Date: 03/11/19 : 00:55:09
400 game(s) loaded
Rank Name Rating Δ + - # Σ Σ% W L D W% =% OppR
---------------------------------------------------------------------------------------------------------
1 McCain X2a 3109 0.0 20 20 400 210.0 52.5 79 59 262 19.8 65.5 3091
2 Stockfish 110319 64 POPC 3091 17.6 20 20 400 190.0 47.5 59 79 262 14.8 65.5 3109
---------------------------------------------------------------------------------------------------------
Time to complete: 00:50:41
]https://github.com/official-stockfish/ ... /pull/2035
As an fyi, I'm working on a standalone mate finder type program (to be named "1st-Mate") which will combine both McCain and Matefinder by Joe Ellis - the source has not yet been released publicly, but when it is released it will never be as fast as Joe Ellis' matefinder in finding mates, but more of a hybrid engine between McCain and Matefinder.
Also, I have updated Joe Ellis' matefinder code with the latest dev-SF code to the best of my abilities ( again you must compile your own exe's). This might be of some interest since Joe's matefinder code has not been updated on GitHub since late December.
https://github.com/MichaelB7/Stockfish/tree/mate_finder
-
- Posts: 229
- Joined: Thu Mar 10, 2016 6:17 pm
- Location: Canada
Re: New engine releases 2019
Thanks for this Mike - I tried compiling and I got one warning and one error - this is just a FYI ..
I'll wait for the official macOS release.
ic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto -c -o search.o search.cpp
search.cpp17: warning: unused function 'futility_move_count'
[-Wunused-function]
constexpr int futility_move_count(bool improving, int depth) {
^
1 warning generated.
g++ -DAdd_Features -DMaverick -march=native -Wall -Wcast-qual -fno-exceptions -std=c++11 -pedantic -Wextra -Wshadow -m64 -arch x86_64 -mmacosx-version-min=10.14 -DNDEBUG -O3 -mdynamic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto -c -o thread.o thread.cpp
g++ -DAdd_Features -DMaverick -march=native -Wall -Wcast-qual -fno-exceptions -std=c++11 -pedantic -Wextra -Wshadow -m64 -arch x86_64 -mmacosx-version-min=10.14 -DNDEBUG -O3 -mdynamic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto -c -o timeman.o timeman.cpp
g++ -DAdd_Features -DMaverick -march=native -Wall -Wcast-qual -fno-exceptions -std=c++11 -pedantic -Wextra -Wshadow -m64 -arch x86_64 -mmacosx-version-min=10.14 -DNDEBUG -O3 -mdynamic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto -c -o tt.o tt.cpp
g++ -DAdd_Features -DMaverick -march=native -Wall -Wcast-qual -fno-exceptions -std=c++11 -pedantic -Wextra -Wshadow -m64 -arch x86_64 -mmacosx-version-min=10.14 -DNDEBUG -O3 -mdynamic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto -c -o uci.o uci.cpp
g++ -DAdd_Features -DMaverick -march=native -Wall -Wcast-qual -fno-exceptions -std=c++11 -pedantic -Wextra -Wshadow -m64 -arch x86_64 -mmacosx-version-min=10.14 -DNDEBUG -O3 -mdynamic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto -c -o ucioption.o ucioption.cpp
g++ -DAdd_Features -DMaverick -march=native -Wall -Wcast-qual -fno-exceptions -std=c++11 -pedantic -Wextra -Wshadow -m64 -arch x86_64 -mmacosx-version-min=10.14 -DNDEBUG -O3 -mdynamic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto -c -o polybook.o polybook.cpp
g++ -DAdd_Features -DMaverick -march=native -Wall -Wcast-qual -fno-exceptions -std=c++11 -pedantic -Wextra -Wshadow -m64 -arch x86_64 -mmacosx-version-min=10.14 -DNDEBUG -O3 -mdynamic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto -c -o syzygy/tbprobe.o syzygy/tbprobe.cpp
g++ -o McCain-X2a benchmark.o bitbase.o bitboard.o endgame.o evaluate.o main.o material.o misc.o movegen.o movepick.o pawns.o position.o psqt.o search.o thread.o timeman.o tt.o uci.o ucioption.o polybook.o syzygy/tbprobe.o -m64 -arch x86_64 -mmacosx-version-min=10.14 -lpthread -DAdd_Features -DMaverick -march=native -Wall -Wcast-qual -fno-exceptions -std=c++11 -pedantic -Wextra -Wshadow -m64 -arch x86_64 -mmacosx-version-min=10.14 -DNDEBUG -O3 -mdynamic-no-pic -DIS_64BIT -msse -msse3 -mpopcnt -DUSE_POPCNT -DUSE_PEXT -mbmi2 -flto
strip McCain-X2a
cp McCain-X2a /Users/michaelbyrne/cluster.mfb
cp: /Users/michaelbyrne/cluster.mfb: No such file or directory
make: *** [build] Error 1
➜ src
-
- Posts: 1643
- Joined: Tue May 20, 2008 4:57 pm
- Location: Augsburg - Germany