Integral 7
https://github.com/aronpetko/integral/releases/tag/v7
LTC selfplay progress:
Elo | 56.93 +- 4.72 (95%)
Conf | 40.0+0.40s Threads=1 Hash=32MB
Games | N: 5000 W: 1627 L: 815 D: 2558
Penta | [2, 274, 1145, 1068, 11]
Integral 6 is world's #11 in my UHO-Top15 Ratinglist. Will Integral 7 make it into the Top10? Result in 4 days... stay tuned!
New engine releases & news H1 2025
Moderator: Ras
-
- Posts: 2669
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
-
- Posts: 1069
- Joined: Thu Mar 09, 2006 4:15 pm
- Location: Long Island, NY, USA
Re: New engine releases & news H1 2025
Hi all,Norm Pollock wrote: ↑Wed Feb 05, 2025 4:31 pm Hi all,
I released a tool "epdFaux" in my 40H-EPD-2025 package of "epd" tools.
"epdFaux" removes all "faux" (false, fake, spurious) en passant notations in an "epd" file.
This includes "faux" notations due to (1) lack of opponent pawn to make the capture; (2) opponent's capturing pawn pinned to its own king which prevents the capture; and (3) the two-space pawn move uncovering a discovered check on the opponent's king preventing an ep capture.
Links below.
Norm Pollock
I recently became aware of some bugs in the above release of "epdFaux". These bugs can affect output. They have been corrected in an update that is now available at the links below. "epdFaux" is in the download for "40H-EPD" tools.
-Norm
-
- Posts: 540
- Joined: Tue Feb 04, 2014 12:25 pm
- Location: Gower, Wales
- Full name: Colin Jenkins
-
- Posts: 2669
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: New engine releases & news H1 2025
Alexandria 8.0
https://github.com/PGG106/Alexandria/re ... tag/v8.0.0
LTC Progression-test, using my UHO openings (Progression test was run on the Pohl.pgn book on OpenBench):
Elo | 42.25 +- 4.53 (95%)
Conf | 60.0+0.60s Threads=1 Hash=64MB
Games | N: 5000 W: 1592 L: 987 D: 2421
Penta | [4, 283, 1327, 876, 10]
https://github.com/PGG106/Alexandria/re ... tag/v8.0.0
LTC Progression-test, using my UHO openings (Progression test was run on the Pohl.pgn book on OpenBench):
Elo | 42.25 +- 4.53 (95%)
Conf | 60.0+0.60s Threads=1 Hash=64MB
Games | N: 5000 W: 1592 L: 987 D: 2421
Penta | [4, 283, 1327, 876, 10]
-
- Posts: 6884
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: New engine releases & news H1 2025
Hi there,
A newer Fire 10.0:
https://github.com/FireFather/fire/tree/main/src
Uralochka Beta 8:
https://gitlab.com/freemanzlat/uralochka3/-/pipelines
Caissa 1.21.7:
https://github.com/Witek902/Caissa/releases
Best
Frank
A newer Fire 10.0:
https://github.com/FireFather/fire/tree/main/src
Uralochka Beta 8:
https://gitlab.com/freemanzlat/uralochka3/-/pipelines
Caissa 1.21.7:
https://github.com/Witek902/Caissa/releases
Best
Frank
-
- Posts: 43676
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
Re: New engine releases & news H1 2025
I don't think that Norm will mind me telling you what he told me - that it hasn't been officially 'released'...just updated. no improvement in Elo, just code changes and 'optimizations'...nnue evaluation is the same as 9.3.Frank Quisinsky wrote: ↑Thu Mar 13, 2025 9:16 am Hi there,
A newer Fire 10.0:
https://github.com/FireFather/fire/tree/main/src
Best
Frank
gbanksnz at gmail.com
-
- Posts: 6884
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: New engine releases & news H1 2025
Hi Graham,
I follow your tourneys.
Thanks for your efforts all the years!!
Fire:
Think so!
Cinder is very interesting, also Inanis ... I think both programs with a great future.
Found Cinder in one of your tounreys.
Best
Frank
I follow your tourneys.
Thanks for your efforts all the years!!
Fire:
Think so!
Cinder is very interesting, also Inanis ... I think both programs with a great future.
Found Cinder in one of your tounreys.
Best
Frank
-
- Posts: 2669
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: New engine releases & news H1 2025
Obsidian 15.09
The author asked me for a new testrun of Obsidian and provided me an avx512 binary. You can download it, too, if you are interested, but you need a CPU, supporting avx512:
The author asked me for a new testrun of Obsidian and provided me an avx512 binary. You can download it, too, if you are interested, but you need a CPU, supporting avx512:
-
- Posts: 460
- Joined: Fri Dec 16, 2016 11:04 am
- Location: France
- Full name: Richard Delorme
Re: New engine releases & news H1 2025
Dumb 2.2 is out: https://github.com/abulmo/Dumb/releases/tag/v2.2
- use magic bitboard for move generation
- retuned evaluation function (still limited to material & positional square table).
- retuned search parameters + probcut added
- much better comments following the D convention
- added logging facilities & eval/board display to better understand how the engine works
This version is bout 20 Elo stronger than version 2.1.
- use magic bitboard for move generation
- retuned evaluation function (still limited to material & positional square table).
- retuned search parameters + probcut added
- much better comments following the D convention
- added logging facilities & eval/board display to better understand how the engine works
This version is bout 20 Elo stronger than version 2.1.
Richard Delorme
-
- Posts: 460
- Joined: Fri Dec 16, 2016 11:04 am
- Location: France
- Full name: Richard Delorme
Re: New engine releases & news H1 2025
Unfortunately I lately introduce a regression before releasing this version.abulmo2 wrote: ↑Wed Mar 19, 2025 10:37 am Dumb 2.2 is out: https://github.com/abulmo/Dumb/releases/tag/v2.2
- use magic bitboard for move generation
- retuned evaluation function (still limited to material & positional square table).
- retuned search parameters + probcut added
- much better comments following the D convention
- added logging facilities & eval/board display to better understand how the engine works
This version is bout 20 Elo stronger than version 2.1.
Version 2.3 fixes it: https://github.com/abulmo/Dumb/releases/tag/v2.3
Richard Delorme