ShashChess

[CornfedForever]

https://github.com/amchess/ShashChess/wiki/Matches
There is another testing strategy based on positional characteristics, I illustrated in the wiki.
NNUE killed someway chess so mostly games are draw.
For the rest, ShashChess has a more realistic evaluation.
Andrea

"Killed"...or is it simply that NNUE raised the level of play for all engines using it, helping to...level the playing field. The engines ARE stronger.
That is to say, better for analysis and 'truth' about positions, but bad for win/loss outcomes among those who simply push engines thru their paces.

[Eduard]

Hello Andrea, I agree that Stockfish is a Bullet Monster. I do not like this. But what should the Stockfish community do? You need a large number of games, and only Bullet can do that.

Before few years I used to like shashchess. But as NNUE continues to make great steps, I see shashchess getting weaker (I'm not talking about testing positions).

Unfortunately, I don't share the opinion that Shashchess plays better than Stockfish in longer games. Before few years I had about 7.000 kns, but now it's 20.000 kns. This results in a transmission of 5 minutes game with 7000 kns at least 13 minutes.

On some losing games from the server (I get databases and study winning games) that Shashchess played, saw that Shashchess plays strange in critical positions. Often not even -1 is shown in a clear losing position. Sometimes Stickfish is already at +3. Shashchess seemed to me like he was blind.

https://github.com/amchess/ShashChess/wiki/Matches
There is another testing strategy based on positional characteristics, I illustrated in the wiki.
NNUE killed someway chess so mostly games are draw.
For the rest, ShashChess has a more realistic evaluation.
Andrea

Strategy or not, I've seen more losses from shashchess on the server than wins. Shashchess often thought they were still on course for a draw even though the position was totally lost. If you see that as realistic, then keep it up. It's not my thing.

[amchess]

ShashChess 25.2
https://github.com/amchess/ShashChess/releases/tag/25.2
ShashChess test match
https://github.com/amchess/ShashChess/wiki/Matches
New 50 positions caos test suite for match play
https://github.com/amchess/ShashChess/t ... tSuite2022

Brainlearn 19.1
https://github.com/amchess/BrainLearn/releases/tag/19.1

[amchess]

https://github.com/amchess/BrainLearn/r ... tag/19.2.1
https://github.com/amchess/ShashChess/releases/tag/25.4

[amchess]

https://github.com/amchess/ShashChess/releases/tag/26
https://github.com/amchess/BrainLearn/releases/tag/20

[peter]

https://github.com/amchess/ShashChess/releases/tag/26
https://github.com/amchess/BrainLearn/releases/tag/20

Thank you, Andrea.
BTW 25.3 is included in ShashChess 26- download too.

[amchess]

https://github.com/amchess/ShashChess/releases/tag/26
https://github.com/amchess/BrainLearn/releases/tag/20

Thank you, Andrea.
BTW 25.3 is included in ShashChess 26- download too.

Corrected.
Thanks,
Andrea

[amchess]

https://github.com/amchess/ShashChess/w ... te-results
https://github.com/amchess/ShashChess/wiki/Matches

[Eduard]

Brainlearn 20 vulkan (64-bit) external network:

Filehorst.de
https://filehorst.de/d/eyaIBycf

Pixeldrain:

https://pixeldrain.com/u/yC9kVhEX

Brainlearn 20 vulkan is a modified Brainlearn 20.1 engine. I made this engine, because I want to test the learning ability of Brainlearn/Shashchess. Andrea Manzo, author of Brainlearn said that the learning properties of Brainlearn are better than those of StockfishMZ/Eman. Also, the learning file does not have to be defragmented. Unfortunately, I'm not satisfied with Shashchess/Brainlearn because I don't like the evaluation (sorry Andrea).

Changes from Brainlearn 20 vulkan to Brainlearn 20.1:

The search has been modified in a few places.
Scaling/Optimism reset to Stockfish dev.
Pieces values were set proportional to PawnValueMg = 100 (Stockfish dev = 126) for dynamic playing style.
MinimumThinkingTime has been implemented. If you play on the server PlayChess, you should set the value = 100. For offline games the value should stay at 0.

Engine-Options:



In my EN-Test 2022 test, Brainlearn 20 vulkan achieved 109 points (109 out of 120) and unofficially took 4th place (only one engine per author makes it into the list). Before that there are only 3 less selective engines that have a small advantage with 60s thinking time in such tests.

ShashChess 25.4 GoldDigger gets 105 solutions, and 104 solutions by Shashchess default. Brainlearn default is weaker than Shashchess.

The Rating List can be seen here (please scroll down). All test positions can be downloaded. Anyone can test for themselves:

https://solistachess.jimdosite.com/testing/

Here are two practical positions:

EN 044 - Mittelspiel Line 0.0, Qxe5
[fen]4q1kr/p6p/1prQPppB/4n3/4P3/2P5/PP2B2P/R5K1 w - - 0 1[/fen]


Analysis by BrainLearn 20 vulkan-x86-64-avx2:

1.Qxe5 fxe5 2.Rf1 Qe7 3.Bd1 Rc4 4.Bb3 b5 5.a4 a6 6.Kg2 g5 7.Rf5 Qg7 8.Bxg7 Kxg7 9.Bxc4 bxc4 10.Kf3 Re8 11.Rxg5+ Kf6 12.Rh5 Rb8 13.e7 Kxe7 14.Rxe5+ Kf6 15.Rf5+ Kg7 16.h4 Rxb2 17.Ke3 Rh2 18.h5 Kh6 19.Ra5 Rxh5 20.Rxa6+ Kg7 21.a5 Rh3+ 22.Kd4 Rd3+ 23.Kxc4 Rd8 24.Rb6 h5 25.a6 h4 26.Rb7+ Kg6 27.a7 Ra8
+- (3.81) Depth: 30/60 00:00:03 71233kN, tb=16106
White is winning

Analysis by BrainLearn 19:

1.Qa3 Rxe6 2.Qxa7 Qe7 3.Qa8+ Qe8 4.Qb7 Qe7 5.Qc8+ Qe8 6.Qb7
= (0.00) Depth: 54/31 00:00:27 483MN, tb=127890
The position is equal

The following position was played live on the server a few days ago. The move Qb5?? is a fatal mistake. I only know a few engines that quickly see that Qb5 is a blunder.

Taktik-Abwehr Line 0.0, Qe7
[fen]r3r2k/p1pq3p/R3bppN/2P5/1p1pP1P1/3P4/2P2P1P/3Q2RK b - - 0 1[/fen]


Analysis by ShashChess 25.4:

30...Qb5 31.Qf3 Kg7 32.Rxe6 Rxe6 33.Nf5+ Kf7 34.Qh3 Kg8 35.Nxd4 Qd7 36.Nxe6 Qxe6 37.f4 a5 38.f5 Qe7 39.Qh6 b3 40.cxb3 g5 41.Re1 Qe5 42.Qh3 Rb8 43.Qg3 Qxc5 44.e5 fxe5 45.Qxe5 Qxe5 46.Rxe5 Rxb3 47.Rxa5 Rxd3 48.Ra8+ Kf7 49.Rc8 Rc3 50.Kg2 Rc2+ 51.Kf3 Kf6 52.h3 Rc3+ 53.Kg2 Rc2+ 54.Kg3 Rc3+
= (0.12 --) Depth: 46/49 00:00:29 604MN, tb=50729
The position is equal

Analysis by BrainLearn 20 vulkan-x86-64-avx2:

30...Qe7 31.Qd2 Qxc5 32.h4 Re7 33.Rga1 Rf8 34.Rxa7 Qc3 35.Qf4 Kg7 36.g5 Bc8 37.Kg1 fxg5 38.Qxg5 Rfe8 39.Ra8 Bd7 40.R1a6 Qxc2 41.Rxe8 Rxe8 42.Rf6 Be6 43.Rf3 b3 44.Nf7 Qd1+ 45.Kg2
+/= (0.43) Depth: 26/49 00:00:02 40201kN, tb=109
White is slightly better

Brainlearn 20 volcano is great!

I'm live on PlayChess, so if you want you can watch the engine play. Or even better: Play and test it yourself. Whether Brainlearn's ability to learn is better than Eman's can only be said after a few thousand played games.

Eduard

[Eduard]

Sorry guys, I compiled and uploaded an older code.

Here is version 20.1:

Filehorst.de

https://filehorst.de/d/egEmqtis

Pixeldrain:

https://pixeldrain.com/u/sURk3hFH