I test stability of deterministic engines for future tournaments.
both clover and plentychess have bugs at 512 nodes per move and lost one of my games by illegal moves.
Engines used the minimal hash they can in the options(2 mbytes or 1 mbyte)
[Round "4"]
[Board "256"]
[White "Clover 6.1 512 nodes"]
[Black "Alexandria-6.1.0 512 nodes"]
[Result "0-1"]
[Termination "illegal move"]
[ECO "D05"]
[Opening "Queen's pawn game"]
[TimeControl "nodes: 8"]
[PlyCount "38"]
1.d4 {+1.0/4 10 216 429/505/66} d5 {+0.2/4 23 334} 2.Nf3 {+0.6/4 10 266 325/585/90} e6 {+0.3/4 10 504}
3.e3 {+0.8/4 10 237 370/561/69} Nf6 {+0.2/3 10 202} 4.c4 {+0.3/4 10 348 225/657/118; D05: Queen's pawn game} c5 {+0.6/3 10 499}
5.a3 {+0.2/4 10 481 206/675/119} Nc6 {+0.3/3 10 420} 6.Nc3 {+0.2/4 10 473 184/694/122} a6 {+0.3/3 10 234}
7.cxd5 {+0.2/4 10 456 177/708/115} exd5 {+0.4/4 10 430} 8.Be2 {+0.8/4 10 321 332/620/48} c4 {-0.6/3 10 343}
9.b3 {+1.5/4 10 183 570/413/17} b5 {-0.1/4 51 447} 10.O-O {+1.1/4 10 252 394/576/30} Bd6 {-1.0/4 12 306}
11.Re1 {+1.7/4 10 443 608/381/11} O-O {-1.9/4 10 447} 12.h3 {-0.1/3 10 197 95/775/130} Bf5 {-2.4/4 27 342}
13.g4 {+4.0/4 10 206 980/20/0} Bg6 {-3.3/4 10 449} 14.g5 {+1.9/4 10 278 680/314/6} Ne4 {-4.4/3 10 384}
15.Nxd5 {-2.1/4 10 295 4/251/745} cxb3 {-4.4/3 23 249} 16.h4 {-0.3/4 10 251 55/804/141} Na5 {-5.3/3 10 342}
17.Bb2 {-2.3/5 10 445 2/210/788} Rc8 {-5.9/4 10 357} 18.Ne5 {-1.9/4 10 272 4/332/664} Rc2 {-7.3/3 10 335}
19.Rb1 {-3.3/3 10 415 0/43/957} Nc4 {-8.6/3 10 350} 0-1
[/pgn]
[pgn]
[Event "Computer event"]
[Site "Somewhere on Earth"]
[Date "2024.04.06"]
[Time "10:33:25"]
[Round "24"]
[Board "528"]
[White "Igel 3.5.0 512 nodes"]
[Black "PlentyChess 1.0.0 512 nodes"]
[Result "1-0"]
[Termination "illegal move"]
[ECO "D04"]
[Opening "Queen's pawn game"]
[TimeControl "nodes: 512"]
[PlyCount "45"]
1.d4 {+1.1/6 10 285} Nf6 {+0.4/4 10 481} 2.Nf3 {+0.9/5 10 244} d5 {+0.4/4 10 326}
3.e3 {+0.5/5 10 306} c5 {+0.1/3 10 252; D04: Queen's pawn game} 4.dxc5 {+2.4/5 10 267} Nc6 {+0.4/3 10 422}
5.c4 {+1.4/5 10 246} e5 {+0.4/3 10 253} 6.cxd5 {+1.4/5 10 417} Qxd5 {+0.5/2 10 119}
7.Qxd5 {+0.5/6 10 369} Nxd5 {+0.1/3 10 202} 8.a3 {+2.2/4 10 356} Bxc5 {-0.4/4 10 449}
9.b4 {-0.1/5 10 234} Be7 {-0.3/3 10 463} 10.Bb2 {+1.8/5 10 315} f6 {-0.3/2 10 316}
11.Bc4 {+0.1/5 10 165} Nb6 {-0.4/4 10 477} 12.Bb3 {+0.3/5 10 290} a5 {-0.6/3 10 221}
13.Ke2 {-1.6/4 10 424} axb4 {-1.3/4 10 428} 14.axb4 {-1.6/5 10 360} Rxa1 {-1.3/4 10 442}
15.Bxa1 {-1.5/5 10 315} Bxb4 {-1.4/4 10 425} 16.Bc3 {-1.1/5 10 336} Ke7 {-1.4/3 10 370}
17.Nfd2 {-1.4/5 10 281} Bd7 {-1.7/3 10 348} 18.Ne4 {-2.1/5 10 433} Rc8 {-1.8/3 10 416}
19.Rd1 {-2.4/4 10 164} f5 {-2.2/3 10 463} 20.Ng5 {-3.2/4 10 253} h6 {-2.1/3 10 342}
21.Nf3 {-3.6/5 10 482} Be6 {-2.0/2 10 429} 22.Nxe5 {-2.2/4 10 508} Bxb3 {-0.9/3 10 433}
23.Rd3 {-6.9/4 10 372} 1-0
[/pgn]
clover and Plentychess bug in 512 nodes per move
Moderator: Ras
-
Uri Blass
- Posts: 11078
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
-
Antihelion
- Posts: 30
- Joined: Tue Mar 26, 2024 8:21 pm
- Full name: Lyndon S. Sears
Re: clover and Plentychess bug in 512 nodes per move
Could you provide more information on the exact illegal move that was played?
-
Uri Blass
- Posts: 11078
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: clover and Plentychess bug in 512 nodes per move
I do not know but I suspect a1-a1 because the engine does not give me output at depth 1.Antihelion wrote: ↑Sat Apr 06, 2024 9:56 am Could you provide more information on the exact illegal move that was played?
I could reproduce the games but the gui does not show me the illegal moves.
Edit in the opening position if I type go nodes 20 then the engine types nothing after best move and the engine exit for PlentyChess.
If I type go nodes 20 for Clover the engine respond best move a1a1
-
lucametehau
- Posts: 128
- Joined: Thu Apr 22, 2021 3:56 pm
- Location: Bucharest, Romania
- Full name: Metehau Luca
Re: clover and Plentychess bug in 512 nodes per move
Indeed it kindof is/was a bug, I think this is currently 99% fixed in the development version. There still are some shady cases, because of how Clover counts nodes.Uri Blass wrote: ↑Sat Apr 06, 2024 11:39 amI do not know but I suspect a1-a1 because the engine does not give me output at depth 1.Antihelion wrote: ↑Sat Apr 06, 2024 9:56 am Could you provide more information on the exact illegal move that was played?
I could reproduce the games but the gui does not show me the illegal moves.
Edit in the opening position if I type go nodes 20 then the engine types nothing after best move and the engine exit for PlentyChess.
If I type go nodes 20 for Clover the engine respond best move a1a1
-
Yoshie2000
- Posts: 1
- Joined: Mon Jan 29, 2024 9:51 am
- Full name: Patrick Leonhardt
Re: clover and Plentychess bug in 512 nodes per move
Hi Uri,Uri Blass wrote: ↑Sat Apr 06, 2024 11:39 amI do not know but I suspect a1-a1 because the engine does not give me output at depth 1.Antihelion wrote: ↑Sat Apr 06, 2024 9:56 am Could you provide more information on the exact illegal move that was played?
I could reproduce the games but the gui does not show me the illegal moves.
Edit in the opening position if I type go nodes 20 then the engine types nothing after best move and the engine exit for PlentyChess.
If I type go nodes 20 for Clover the engine respond best move a1a1
I have fixed this issue in the current development versions. The next release will no longer play illegal moves if it can't finish a full depth 1 search.