Stockfish 15.1 is ready

[mehmet123]

Stockfish 15.1 was released after 8 months after Stockfish 15.
According to the regression tests elo difference between Stockfish 15.1 and Stockfish 15 isn't so much, but the difficulty of developing such a super strong chess engine should not be forgotten.
https://stockfishchess.org/download/
https://abrok.eu/stockfish/



(SF SkyNet 15.1:
https://www.mediafire.com/file/s94xh3mo ... 1.zip/file
The only difference is polyglot book support added to Stockfish 15.1)

[CornfedForever]

Stockfish 15.1 was released after 8 months after Stockfish 15.
According to the regression tests elo difference between Stockfish 15.1 and Stockfish 15 isn't so much, but the difficulty of developing such a super strong chess engine should not be forgotten.
https://stockfishchess.org/download/
https://abrok.eu/stockfish/



(SF SkyNet 15.1:
https://www.mediafire.com/file/s94xh3mo ... 1.zip/file
The only difference is polyglot book support added to Stockfish 15.1)

Any idea how it differs from Fridays development version?

[sarona]

There is no change other than line 71 in misc.cpp in the abrok releases

Author: Joost VandeVondele
Date: Fri Dec 2 20:06:59 2022 +0100
Timestamp: 1670008019

const string version = "dev";
--------------------------------------------------------------------
Author: Joost VandeVondele
Date: Sun Dec 4 14:17:15 2022 +0100
Timestamp: 1670159835

const string version = "15.1";


No other code changes.

[Vinvin]

Progress for SF 15.1 over SF 15.0 :

Each test : 60'000 games, Time Control = 60s+0.6s, 1 thread.

+17.04 : Book UHO (unbalanced-human-openings)
+53.51 : Double Fischer Random Chess
+29.45 : Fischer Random Chess
+7.46 : Standard Chess, Standard Book


From : https://tests.stockfishchess.org/tests/ ... ltc_only=1

[CornfedForever]

There is no change other than line 71 in misc.cpp in the abrok releases

Author: Joost VandeVondele
Date: Fri Dec 2 20:06:59 2022 +0100
Timestamp: 1670008019

const string version = "dev";
--------------------------------------------------------------------
Author: Joost VandeVondele
Date: Sun Dec 4 14:17:15 2022 +0100
Timestamp: 1670159835

const string version = "15.1";


No other code changes.

Thanks!

[Jouni]

This string change was -8 at NCM .

[mehmet123]

Progress for SF 15.1 over SF 15.0 :

Each test : 60'000 games, Time Control = 60s+0.6s, 1 thread.

+17.04 : Book UHO (unbalanced-human-openings)
+53.51 : Double Fischer Random Chess
+29.45 : Fischer Random Chess
+7.46 : Standard Chess, Standard Book


From : https://tests.stockfishchess.org/tests/ ... ltc_only=1

And regression is +5.97 elo at VLTC (8 * 30 sec + 0.3 sec).
+6 elo progress at 8 months isnt hopeful. I think elo progress at rapid and long time control is at the range of 0 - 5 elo. There isnt any new NNUE net for 5 months.

[Eduard]

Nice! But I don't need this engine. I'm orienting myself towards Stockfish dev, and my engines as well as Vulkan contain all Stockfish patches up to now.

[Jouni]

No wonder, if minimal progress in standard chess. SF NNUE has "practically" solved chess. It's draw . But you can detect draw a little faster in future?

[Eduard]

But there are also tournaments where one side has a good advantage! TCEC for example. Or now on InfinityChess similar to TCEC. I don't want to despise +8 Elo!