Chess System Tal 2 NNUE has been released

[mclane]

The problems stockfish has, especially with consecutive x in y time controls without increment go far deeper then move overhead problems in arena.

It's not even a sudden death time control.
It's a relatively easy x in y time control.
So no rocket science needed.

Thorsten, you already have proven in CSS that you are just too dumb for a proper setup even on a modern computer. So lets finsih this discussion. You won't learn anything.

Let's hope the stockfish makers are not that ignorant like you.

[mclane]

very strong performance on the 8 core too:

Code: Select all

    Motor                                   Punkte    St   CS   CS   Et   Dr   Re   Dr   Be   Lc   De   Bo   Wa   Ko   Pe   Ch   De   Eq   De   CS   Co    S-B
01: Stockfish16 [t4, mo1000]                58,5/76 ···· ==== 1=1= =1== ==== ==== ==== ==1= 1=== ===1 ==1= 1111 =111 1111 1=11 1111 1111 1111 1111 1111  1939,5
02: CSTal2-v21-E1019-Elo-avx2 [t4]          57,5/76 ==== ···· ==== ==== 01== ==== ==== ===1 ==1= 1111 101= 111= 111= 1111 =111 11=1 1111 1111 1111 1111  1887,7
03: CSTal2-v21-E1162-130-EAS-avx2 [t4]      56,5/76 0=0= ==== ···· ==== ==== =1== 11== 1=0= ===1 111= 1111 11== =111 1==1 ==11 1111 1=11 1111 1111 1111  1858,5
04: Ethereal-14.00-pext-avx2 [t4]           54,5/76 =0== ==== ==== ···· 01== ==== ==== ==== =1== =1== =1== 111= 1111 1111 1110 1111 0111 1111 1111 1111  1777,5
05: Dragon-3.1-64bit-avx2 [t4]              54,5/76 ==== 10== ==== 10== ···· ==== ==0= ==01 1=11 ==11 1=== 1=10 1==1 1101 1111 1111 =111 1111 1111 1111  1777,2
06: Rebel-16.2 [t4]                         54,5/76 ==== ==== =0== ==== ==== ···· 0=01 ===1 ==== 1==1 1=== 1==1 1=11 1==1 1111 1111 1111 1111 1111 1111  1746,5
07: Dragon-3.1-64bit-avx2 [t4, mcts]        52,5/76 ==== ==== 00== ==== ==1= 1=10 ···· ==== =100 =011 11=1 1111 ===1 ===1 11=1 1==1 =111 1111 1111 1111  1715,0
08: Berserk-11-x64-avx2-pext [t4]           50,0/76 ==0= ===0 0=1= ==== ==10 ===0 ==== ···· ==== =0== 1=== 1=== =1=1 1=1= 1111 1=11 1111 11=1 1111 1111  1588,7
09: Lc0 0.29.0 rc1 radeon gpu dx12 [753723] 48,0/76 0=== ==0= ===0 =0== 0=00 ==== =011 ==== ···· ==11 10=1 =11= 110= 1=== 1=11 1=1= =111 1=11 1111 1111  1520,0
10: DeepHiarcs15.0 [t4]                     39,0/76 ===0 0000 000= =0== ==00 0==0 =100 =1== ==00 ···· ===1 =1=1 011= =1== =1=0 ==== 111= =111 =1=1 1111  1184,2
11: Booot7 [t4]                             37,0/76 ==0= 010= 0000 =0== 0=== 0=== 00=0 0=== 01=0 ===0 ···· ==== 00== =1=1 1=00 ===1 1=11 =111 1111 1=11  1122,0
12: Wasp600-win-avx [t4]                    36,5/76 0000 000= 00== 000= 0=01 0==0 0000 0=== =00= =0=0 ==== ···· ==== =0== 111= 1111 11=1 1111 1111 1111  954,25
13: Komodo-14.1-64bit [t4, mcts]            34,0/76 =000 000= =000 0000 0==0 0=00 ===0 =0=0 001= 100= 11== ==== ···· ==== 11== 11== 1==1 0111 101= 1111  955,75
14: Pedone3.1_avx2 [t4]                     34,0/76 0000 0000 0==0 0000 0010 0==0 ===0 0=0= 0=== =0== =0=0 =1== ==== ···· 1=1= =1== 111= 1=11 11=1 1111  915,50
15: Chiron-5-64bit [t4]                     24,5/76 0=00 =000 ==00 0001 0000 0000 00=0 0000 0=00 =0=1 0=11 000= 00== 0=0= ···· 1==0 =01= 1==1 ==1= 1111  640,00
16: DeepShredder13UCIx64 [t4, buffer6]      20,5/76 0000 00=0 0000 0000 0000 0000 0==0 0=00 0=0= ==== ===0 0000 00== =0== 0==1 ···· 10=1 =0=0 1==1 1101  528,50
17: Equinox 3.30-x64 [t4]                   18,5/76 0000 0000 0=00 1000 =000 0000 =000 0000 =000 000= 0=00 00=0 0==0 000= =10= 01=0 ···· 10== 1=1= 1111  425,25
18: Deep Rybka 4.1 x64 [t4]                 14,0/76 0000 0000 0000 0000 0000 0000 0000 00=0 0=00 =000 =000 0000 1000 0=00 0==0 =1=1 01== ···· =001 0111  289,50
19: CSTal3 5'                               12,0/76 0000 0000 0000 0000 0000 0000 0000 0000 0000 =0=0 0000 0000 010= 00=0 ==0= 0==0 0=0= =110 ···· 0111  228,25
20: Colossus2022a [t4]                      3,5/76  0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0=00 0000 0000 0000 0000 0010 0000 1000 1000 ····   65,00

760 Partien gespielt / Turnier beendet
Name des Turniers: Ryzen7 5700
Ort/ Land: LAPTOP-JTQQP8M7, Deutschland
Spielstufe: Turnier 40/5
Hardware: AMD Ryzen 7 5700U with Radeon Graphics  mit 4.096 MB Speicher

[Nordlandia]

Which executable is recommended for sacrifical play

SmashChess Petrosian vs. Tal

[reflectionofpower]

EAS version

[mclane]

You don't understand anything, otherwise you wouldn't write something like this. You live in the Stone Age of computer chess. You've probably never played chess online (there's not just the Fritz server that you don't like because it's from ChessBase), otherwise you wouldn't write like this. On game servers move overhead has to be set much higher because of LAG. This is the case everywhere where online games are played (also on Lichess). Otherwise your engine will lose almost every online game longer than 80 moves. Any 5-year-old learning chess online will know what to do. But the super expert Thorsten, who has been with computer chess since the beginning, doesn't know that. Isn't that funny!

I have worked with many chess programmers. I never cared how strong the engines were. If they were newcomers or old professionals, I decided for the most interesting job. No matter how big the elo was. Today I see people post ELO values without showing us the Games. As if computer chess is about a 4 digit ELO rating and not about chess content.

E. G. when I contacted Mark Uniacke his Hiarcs was not the strong engine it later was. Or when I worked with Christophe Theron Chess Tiger was not leading the rating lists. But there was a potential in these engines.
When I contacted Chris W. it was similar. His engine was not a leading top 10 chess engine but a very interesting chess engine.

It's imo not important if an engine is 1400 ELO or 2400 ELO or 3400 ELO.
Important is the idea behind it. The potential that is in the idea behind.

Chris W. has shown that CSTal is capable to play strong AND play with sacrifices. He changed the engine from single cpu 32 bit into 64 bit multi processing and NNUE usage.
He began computer chess on Sinclair ZX81 and later for the Sinclair ZX Spectrum. Then Amstrad Joyce (CPM). Then Atari ST and PC.
And always he had the same ideology in mind. An intelligent system that plays good AND Interesting games. Not a bean counter. Not a boring machine chess engine.
Yes I live in the stone age. Because I do look into the games and enjoy them. And I like to see progress bering made, no matter if the engine began with 1400 ELO, 2400 elo or 3400 elo.

That's why I do not do online chess on chess base servers with engines. Because this is not about chess content but ELO.

Congratulations to Chris Whittington and Ed Schröder for bringing out a unique engine on such a high level of computerchess.

The games are often very interesting.
Today Chris engine is within the top 5 leading engines.

[Eduard]

Thorsten, I can only wonder. You tell endless stories from your past, but have absolutely no idea about testing.

I just read a sentence from you in the CSS forum

You probably always have to play increment games because otherwise Stockfish would always lose on time.


No!!

Oh man, I play on all servers and with Stockfish exclusively without increment. We mostly play 5+0. Anyone who knows and reads your sentence can only come to the conclusion that you have no idea. Because you have no idea! There is no more clueless tester than you.

[marsell]

Thorsten, I can only wonder. You tell endless stories from your past, but have absolutely no idea about testing.

I just read a sentence from you in the CSS forum

You probably always have to play increment games because otherwise Stockfish would always lose on time.


No!!

Oh man, I play on all servers and with Stockfish exclusively without increment. We mostly play 5+0. Anyone who knows and reads your sentence can only come to the conclusion that you have no idea. Because you have no idea! There is no more clueless tester than you.

Of course, Stockfish loses without move overhead adjustment in games without increment. Thorsten probably played his tournament with default settings.
One question Eduard: Do you increase move overhead when playing on the game server or not? Maybe you even play with adjustment of all three uci parameters like minimum thinking time, slowmover and move overhead.
Not with Stockfish, but with a clone that has these parameters. And why do you think it has them?
Thorsten played with default settings and those are the worst settings for games without increment. Is this really the tester's fault?
Not for me

[Eduard]

These parameters are there because they need to be customized for each platform. With InfinityChess I have to set MoveOverhead much higher than with PlayChess or Lichess or offline.

I haven't installed ChessSystem Tal yet. Actually, I wanted to test Tal.

But: without Ponder it makes no sense to play on servers. It's amazing that CSTal can't ponder, even though this feature was already there in MS-DOS days. And of course I would have to significantly increase the move overhead on the servers, otherwise CSTal would lose almost every game. I have to intervene and optimize manually. This is normal and should be done everywhere.

[marsell]

I haven't installed ChessSystem Tal yet.

but you philosophize about the ponder problem?
And of course CSTal can ponder
bye

[mclane]

These parameters are there

You can play on chess servers against other people with chess engines. That's your kind of computerchess.
I prefer different setting.