Koivisto 9.2 available

[Jouni]

Rewriting movegen! https://ipmanchess.yolasite.com/compiles.php. Great improvement in test suites: Arasan suite now 159 Koivisto 9 only 150.

[Chessqueen]

Rewriting movegen! https://ipmanchess.yolasite.com/compiles.php. Great improvement in test suites: Arasan suite now 159 Koivisto 9 only 150.

Sorry to ask Sir, but I can NOT find the executable file of version 9.2 as it is here for the previous version 9.0 https://github.com/Luecx/Koivisto/releases/tag/v9.0

[peter]

Great improvement in test suites: Arasan suite now 159 Koivisto 9 only 150.

Thanks for the info, Jouni!

Arasan-suite is an "easy" one (as for hardware- TC to be run with), would compare it to the 594 "hardest" positions out of STS, those are the ones, that can be used as single best move- positions more or less too. With which (hardware-) TC did you let Koivisto run to get "only" 75% of the 200 positions solved?

Didn't make it to prove progress in tactical suites with harder positions and longer TC statistically, but at least a little bit more scoring is counted in Mosca's LAN_v5 set of MEA- STS. So I started a very short TC- match, which is still running, reason for posting about is this cutechess cli- outpout:

Code: Select all

Started game 5 of 100 (Koivisto9 vs Koivsto9.2)
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Bd6 Bxd6 Qxd6 Nxh8 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV:  e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move e8g8 from Koivsto9.2 (1)
Warning: PV: Nd4 Qd2 Qe5+ Kf1 e8g8
Warning: Illegal PV move h8d8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 h8d8
Warning: Illegal PV move h8d8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 h8d8
Warning: Illegal PV move h8f8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 h8f8
Warning: Illegal PV move h8f8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 h8f8
Warning: Illegal PV move h8f8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 c4 dxc4 Rb8 b3 c5 h4 h8f8
Warning: Illegal PV move h8d8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 c4 c3 Nf5 Nd2 cxd3 h4 Qe2+ Qxe2 dxe2+ Ke1 h8d8
Warning: Illegal PV move h8d8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 c4 c3 Nf5 Nd2 cxd3 Nf3 Qd5 c4 Qd6 g4 Ne7 Kg2 h8d8
Warning: Illegal PV move h8d8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 c4 c3 Nf5 Nd2 cxd3 h4 Rb8 Kg1 Rxb2 Nc4 Qxc3 Nxb2 Qxb2 h5 Qf6 Rh3 h8d8
Warning: Illegal PV move h8d8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 c4 c3 Nf5 dxc4 Qf4 Qe2 h8d8
Warning: Illegal PV move h8d8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 c4 dxc4 Qf4 c3 Nf5 Qe2 h8d8
Warning: Illegal PV move h8d8 from Koivsto9.2 (1)
Warning: PV: Kxf7 Qh5+ g6 Qd1 Qe5+ Kf1 c4 c3 Nf5 dxc4 Rb8 b3 Qf6 Qe2 h8d8
Finished game 5 (Koivisto9 vs Koivsto9.2): 1-0 {White wins by adjudication}
Score of Koivisto9 vs Koivsto9.2: 1 - 1 - 2  [0.500] 4

Not sure about the position out of the .pgn, illegal move in pv didn't come to board, game ended normally, probably output was at something around move nr.10.

[pgn][Event "?"]
[Site "?"]
[Date "2023.02.07"]
[Round "3"]
[White "Koivisto9"]
[Black "Koivsto9.2"]
[Result "1-0"]
[EventDate "????.??.??"]
[ECO "B30"]
[PlyCount "48"]
[GameDuration "00:00:23"]
[GameEndTime "2023-02-07T12:33:05.690 Mitteleuropäische Zeit"]
[GameStartTime "2023-02-07T12:32:41.948 Mitteleuropäische Zeit"]
[Opening "Sicilian"]
[Termination "adjudication"]
[TimeControl "15+0.25"]
[Variation "Nimzovich-Rossolimo attack (without ...d6)"]

1.e4 {book} 1...c5 {book} 2.Nf3 {book} 2...Nc6 {book} 3.Bb5 {book} 3...e6 {book} 4.Bxc6 {book} 4...bxc6 {book} 5.d3 {book} 5...d6 {book} 6.e5 {book} 6...dxe5 {book} 7.Nxe5 {+1.47/21 0.49s} 7...Qc7 {-1.23/25 2.7s} 8.Qe2 {+1.16/20 0.39s} 8...Ne7 {-1.05/22 0.40s} 9.Bf4 {+1.39/24 0.78s} 9...Nf5 {-1.10/23 0.44s} 10.Ng6 {+3.89/23 0.31s} 10...Bd6 {+0.98/23 0.62s} 11.Bxd6 {+5.95/22 0.43s} 11...Qxd6 {+1.05/22 0.26s} 12.Nxh8 {+6.84/22 0.46s} 12...Nd4 {+8.19/22 0.31s} 13.Nxf7 {+7.30/21 0.60s} 13...Kxf7 {+5.48/22 0.33s} 14.Qh5+ {+7.41/20 0.34s} 14...Kf8 {-7.39/21 1.2s} 15.Na3 {+7.58/19 0.30s} 15...Qd5 {-7.17/25 1.5s} 16.Qxh7 {+7.68/22 0.28s} 16...c4 {-7.59/25 1.3s} 17.O-O {+7.94/22 0.28s} 17...Ne2+ {-7.72/24 0.76s} 18.Kh1 {-M14/1 0.002s} 18...cxd3 {-7.59/23 0.63s} 19.Qxd3 {+7.68/22 0.43s} 19...Qxd3 {-7.83/21 0.35s} 20.cxd3 {+7.91/23 0.37s} 20...c5 {-7.84/25 1.9s} 21.Nc4 {+8.35/19 0.37s} 21...Bb7 {-7.93/22 1.0s} 22.f3 {+8.49/19 0.33s} 22...Rd8 {-8.11/22 0.61s} 23.Rfe1 {+8.75/22 0.99s} 23...Nf4 {-8.21/21 0.64s} 24.Ne5 {+8.52/21 0.42s} 24...Nxd3 {-8.48/24 1.2s, White wins by adjudication} 1-0
[/pgn]

Can't reproduce the output so far, not wanting to stop the running match, here's the momentary score, TC was 15" for the game and 0.25" per move increment, 8 threads of 16x3.5GHz Ryzen- CPU, avx2-pgo-pext-compile.

Score of Koivisto9 vs Koivsto9.2: 10 - 13 - 57 [0.481] 80
Elo difference: -13.0 +/- 40.9, LOS: 26.6 %, DrawRatio: 71.3 %

So probably no way to get out of error bar <2000 games of even so short TC, guess I will stop already after 100 games, that's why I like to have positional testing also. Not to replace eng-eng-game playing of course, but to get additional information (as Ed Schröder writes on his MEA- site too) out of a minimum of hardware- time.

Edit: 100 games already, stopped there:

Score of Koivisto9 vs Koivsto9.2: 14 - 18 - 68 [0.480] 100
Elo difference: -13.9 +/- 38.6, LOS: 24.0 %, DrawRatio: 68.0 %

[peter]

Can't reproduce the output so far, not wanting to stop the running match,

So as for the output in infinite analysis at Black's move nr.10 now again:

Code: Select all

r1b1kb1r/p1q2ppp/2p1p1N1/2p2n2/5B2/3P4/PPP1QPPP/RN2K2R b KQkq - 0 1

Analysis by Koivisto 9.2:

10...Db7 
  Weiß steht besser: +/- (1.29)  Tiefe: 1/19   00:00:02  118kN
10...Db7 11.Sxh8 
  Weiß steht deutlich besser: +/- (1.51)  Tiefe: 2/19   00:00:02  121kN
...
10...Db7 11.Sxh8 Sd4 12.Dd1 Dxb2 13.Dh5 Le7 14.Sxf7 g6 15.Dxh7 Dxa1 16.Dg8+ Kd7 17.Se5+ 
  Weiß hat entscheidenden Vorteil: +- (3.04)  Tiefe: 13/30   00:00:02  491kN
10...Db7 11.Sxh8 Sd4 12.Dd1 Dxb2 13.Dh5 Le7 14.Sxf7 g6 15.Dxh7 Dxa1 16.Dg8+ Kd7 17.Se5+ 
  Weiß hat entscheidenden Vorteil: +- (3.01)  Tiefe: 14/32   00:00:02  727kN
10...Db7 11.Sxh8 Sd4 12.Dd1 Dxb2 13.Dh5 Sxc2+ 14.Kd1 Df6 15.Kxc2 Dxf4 16.Dxf7+ Dxf7 17.Sxf7 Kxf7 18.Sd2 La6 19.The1 Td8 20.Te3 Le7 21.Tae1 Lf6 22.Sc4 Lb5 23.Txe6 
  Weiß steht auf Gewinn: +- (3.70)  Tiefe: 15/33   00:00:02  1414kN
...
10...Db7 11.Sxh8 Dxb2 12.Le5 Dc1+ 13.Dd1 Dxd1+ 14.Kxd1 f6 15.Lc7 c4 16.dxc4 Lc5 17.Sd2 Lxf2 18.Se4 Ld4 19.c3 Lb6 20.Lxb6 axb6 21.Te1 Ke7 22.Tb1 b5 23.cxb5 cxb5 24.Txb5 Txa2 25.Tc5 La6 26.c4 Kd7 27.Sf7 Sd4 28.Te3 e5 29.Td5+ Kc7 30.Tg3 
  Weiß steht klar auf Gewinn: +- (4.26)  Tiefe: 25/61   00:00:04  76064kN, tb=4
10...Db7 11.Sxh8 Dxb2 12.Le5 Dc1+ 13.Dd1 Dxd1+ 14.Kxd1 f6 15.Lc7 c4 16.dxc4 Lc5 17.Sd2 Lxf2 18.Se4 Ld4 19.c3 Lb6 20.Lxb6 axb6 21.Te1 Ke7 22.Tb1 b5 23.cxb5 cxb5 24.Txb5 Txa2 25.Tb8 Ta1+ 26.Ke2 La6+ 27.Kd2 Ta2+ 28.Kc1 e5 29.g4 
  Weiß steht auf Gewinn: +- (4.20)  Tiefe: 26/61   00:00:04  81440kN, tb=4
...
10...Db6 11.Sxh8 Dxb2 12.Le5 Dc1+ 13.Dd1 Dxd1+ 14.Kxd1 f6 15.Lc7 Kd7 16.La5 La6 17.Sd2 Sd6 18.c4 Le7 19.Sb3 Sb7 20.Ld2 Txh8 21.f4 Ld6 22.Kc2 h5 23.g3 Ke8 24.Tab1 Kf7 25.Le3 e5 26.Sd2 exf4 27.Lxf4 Le7 28.The1 
  Weiß steht auf Gewinn: +- (3.93)  Tiefe: 32/64   00:00:08  267MN, tb=13
10...Ld6 11.Lxd6 Dxd6 12.Se5 Sd4 13.Sc4 Dd5 14.Se3 Sxe2 15.Sxd5 cxd5 16.Kxe2 Ke7 17.b3 Ld7 18.Sc3 h6 19.f4 Lc6 20.Thf1 a5 21.Kd2 h5 22.Tae1 Kd6 23.a4 h4 24.Te5 Tab8 25.Tg5 g6 26.Tg4 Tb7 27.Tg5 d4 28.Se4+ Lxe4 29.dxe4 Kc6 30.Kc1 Tb4 
  Schwarz steht besser: -/+ (-0.86)  Tiefe: 33/64   00:00:40  1710MN, tb=186
10...Ld6 11.Lxd6 Dxd6 12.Se5 Sd4 13.Sc4 Dd5 14.Se3 Sxe2 15.Sxd5 cxd5 16.Kxe2 a5 17.b3 Ld7 18.Sc3 Ke7 19.h4 Thb8 20.Kd2 Lc6 21.Thg1 a4 22.h5 Kd6 23.f4 g5 24.fxg5 Tg8 25.Taf1 Ke7 26.g6 hxg6 27.g4 Th8 28.hxg6 Th2+ 29.Kc1 fxg6 30.Th1 Txh1 31.Txh1 d4 32.Th7+ Kd6 33.Se2 g5 
  Schwarz steht besser: -/+ (-0.89)  Tiefe: 34/83   00:01:24  3843MN, tb=2728
...
10...Ld6 11.Lxd6 Dxd6 12.Se5 Sd4 13.Sc4 Dd5 14.Se3 Sxe2 15.Sxd5 cxd5 16.Kxe2 Ld7 17.h4 Ke7 18.Kd2 a5 19.Sc3 Lc6 20.b3 Thb8 21.f4 a4 22.Tab1 h5 23.The1 Kd6 24.g3 Tb7 25.Te5 g6 26.Tee1 Ke7 27.Sd1 Kd7 28.Sc3 d4 29.Se4 Lxe4 30.Txe4 Tab8 31.Te5 Kd6 32.Te4 Kd5 33.Ta1 Kd6 
  Schwarz steht besser: -/+ (-0.87)  Tiefe: 39/83   00:02:41  7478MN, tb=11846

Suspicious change of mind as for White's or Black's advantage, then the engine crashed (just stopping with GUI- error message "this is a problem in the engine", could be restarted without new start of GUI), then I went into the output- line a little, soon the engine is seeing the 12.Ne5 as as crucial blunder.
After this one 12.Ne5 (?) in outpout quoted above :


r1b1k2r/p4ppp/2pqp3/2p1Nn2/8/3P4/PPP1QPPP/RN2K2R b KQkq - 0 1

Analysis by Koivisto 9.2:

12...Sd4 13.Sc4 Dd5 14.Se3 Sxe2 15.Sxd5 cxd5 16.Kxe2 Ld7 17.Sc3 a5 18.b3 Ke7 19.h4 Lc6 20.Kd2 Thb8 21.h5 f5 22.d4 cxd4 23.Se2 e5 24.f4 Lb5 25.fxe5 Lxe2 26.Kxe2 Tc8 27.Kd2 Ke6 28.Th4 Tc3 29.Txd4 Tac8 30.Tc1 Tg3 31.Tg1 g5 32.hxg6 hxg6 33.Td3 Tg4 34.g3 Th8 35.a4 d4 36.c3 Th2+ 37.Kc1 dxc3 38.Txc3 Kxe5 39.Tc5+ Ke4 40.Txa5
Schwarz steht besser: -/+ (-0.92) Tiefe: 35/70 00:00:39 1901MN, tb=1172

And one ply backwards:

r1b1k2r/p4ppp/2pqp1N1/2p2n2/8/3P4/PPP1QPPP/RN2K2R w KQkq - 0 1

Analysis by Koivisto 9.2:

12.Sxh8 Sd4 13.Sxf7 Dd5 14.De4 Kxf7 15.Sa3 La6 16.f3 Kg8 17.0-0 Dxe4 18.fxe4 e5 19.Tf2 h6 20.Td1 Te8 21.b3 Lb5 22.Tc1 a5 23.h3 a4 24.Sb1 g5 25.Sd2 a3 26.Sf1 La6 27.Se3 Lc8 28.Sc4 Le6 29.Sxe5 Lxh3 30.c3 Sb5 31.gxh3 Txe5 32.Tf6 Sc7 33.Txc6 Se6 34.Kf2 Sf4
Weiß steht klar auf Gewinn: +- (7.68) Tiefe: 37/76 00:03:25 9116MN, tb=24981

Unclear, why Koivisto doesn't take the rook at12th at first in output at postion 2 moves earlier.
Maybe yet some problem with the position within the engine('s compile) itself?

[Jouni]

More results:

Code: Select all

                       
1   Koivisto9.2    +9  +40/=330/-30 51.25%  205.0/400
2   Koivisto9      -9  +30/=330/-40 48.75%  195.0/400

ERET suite with 15s and 6 cores version 9.2 got 89/111 but 9 only 83/111

[peter]

More results:

Code: Select all

                       
1   Koivisto9.2    +9  +40/=330/-30 51.25%  205.0/400
2   Koivisto9      -9  +30/=330/-40 48.75%  195.0/400

ERET suite with 15s and 6 cores version 9.2 got 89/111 but 9 only 83/111

Walter Eigenmann himself would give only 2"/pos. for Eret-suite nowadays:

https://forum.computerschach.de/cgi-bin ... #pid161025

Natürlich ist der ERET nun (im NN-Zeitalter) nicht mehr State of the Art - interessant wäre allenfalls noch eine Testzeit von z.B. 2sec./Zug

Here's the part of the EloStatTS- list of the 128 hard ones

https://www.dropbox.com/s/804b7chwli13laf/1284.epd?dl=0

, that I use with 15"/move and 30 threads of the 16x3.5GHz CPU, 3070ti for LC0

Code: Select all

    Program                                    Elo   +/-  Matches  Score   Av.Op.   S.Pos.   MST1    MST2   RIndex

 15 Stockfishdev-20230104-9fe9ff00-MV4       : 3527    7   2872    54.4 %   3496    66/128    4.1s    9.4s   0.44
 16 Sting28                                  : 3499    8   2790    49.6 %   3502    52/128    3.9s   10.5s   0.32
 17 Ceres0.97RC3-784038                      : 3496    8   2666    48.8 %   3504    49/128    3.3s   10.5s   0.36
 18 Fire8.NN.MC.3-ninu-0.3.bin-MV4           : 3492    7   2632    48.4 %   3503    50/128    4.0s   10.7s   0.35
 19 Ceres0.97RC3-784822                      : 3489    8   2648    47.8 %   3504    48/128    3.5s   10.7s   0.33
 20 Stockfish310722                          : 3489    7   2732    48.3 %   3500    55/128    5.5s   10.9s   0.30
 21 Lc0v0.30.0-dag+git.c91bf77-784038        : 3483    8   2636    47.0 %   3504    45/128    3.5s   11.0s   0.32
 22 Sting28-MV4                              : 3481    8   2627    46.6 %   3504    45/128    3.3s   10.9s   0.30
 23 Lc0v0.30.0-dag+git.c91bf77-784822        : 3474    8   2632    45.6 %   3505    46/128    4.7s   11.3s   0.27
 24 Dragon3.1byKomodoChess-MV4               : 3473    7   2581    45.7 %   3503    47/128    4.7s   11.2s   0.28
 25 SlowChessBlitz2.9avx2-MV4                : 3454    7   2497    42.8 %   3505    43/128    5.5s   11.8s   0.23
 26 Dragon3.1byKomodoChess                   : 3447    8   2433    41.5 %   3506    39/128    5.0s   12.0s   0.25
 27 Minic3.30-forceNNUE-MV4                  : 3444    8   2450    41.1 %   3506    40/128    5.7s   12.1s   0.21
 28 Dragon3.1byKomodoChess-MCTS-MV30         : 3441    8   2448    40.5 %   3508    37/128    4.0s   11.8s   0.13
 29 Ethereal13.75(NNUE)-AnalysisMode-MV4     : 3434    8   2375    39.5 %   3509    34/128    4.3s   12.2s   0.21
 30 Berserk20220725-MV4                      : 3430    8   2356    39.0 %   3508    35/128    4.8s   12.2s   0.20
 31 RubiChess20220813(bmi2)-MV4              : 3420    8   2346    37.4 %   3509    29/128    3.4s   12.4s   0.18
 32 PowerFritz18AVX20-MV4                    : 3403    8   2273    35.0 %   3511    30/128    6.1s   12.9s   0.14
 33 Koivisto9.2                              : 3377    8   2193    31.3 %   3514    21/128    5.0s   13.4s   0.12
 34 Revenge3.0b-MV4                          : 3373    8   2346    31.2 %   3511    30/128   12.4s   14.4s   0.03
 35 Koivisto8.16                             : 3371    8   2147    30.4 %   3514    19/128    3.5s   13.3s   0.12
 36 Koivisto9.0                              : 3371    9   2155    30.4 %   3515    18/128    3.4s   13.4s   0.14

MST1  : Mean solution time (solved positions only)
MST2  : Mean solution time (solved and unsolved positions)
RIndex: Score according to solution time ranking for each position

Ferdinand Mosca's LAN_v5.epd of STS from here:

https://github.com/fsmosca/STS-Rating/tree/master/epd



LC0 needs so much more time because of longer initialisation- time on one hand and even more than this, MEA- output at end of runs shows that it uses only (rather exactly) half of the given time, so to have it use 500 msec you have to type 1000 instead, regards

[Plutie]

...

LC0 needs so much more time because of longer initialisation- time on one hand and even more than this, MEA- output at end of runs shows that it uses only (rather exactly) half of the given time, so to have it use 500 msec you have to type 1000 instead, regards

if you want to force lc0 to use all its allocated time, you can set SmartPruningFactor to 0, and lower MoveOverheadMs to 1.

[peter]

...

LC0 needs so much more time because of longer initialisation- time on one hand and even more than this, MEA- output at end of runs shows that it uses only (rather exactly) half of the given time, so to have it use 500 msec you have to type 1000 instead, regards

if you want to force lc0 to use all its allocated time, you can set SmartPruningFactor to 0, and lower MoveOverheadMs to 1.

Thanks for the info!

[Luecx]

Thanks for posting this. I wasnt aware of this bug. When rewriting move-generation, I made sure the bench doesnt change on a few test positions. The issue only seems to arise when using normal tournament mode. when using fixed depth searches, this problem didnt occure. Anyway, I pushed a quick fix.

Greetings
Finn

[peter]

Thanks for posting this. I wasnt aware of this bug. When rewriting move-generation, I made sure the bench doesnt change on a few test positions. The issue only seems to arise when using normal tournament mode. when using fixed depth searches, this problem didnt occure. Anyway, I pushed a quick fix.

Thanks a lot, Finn, and thanks to Ipman for the already new compiles too

https://ipmanchess.yolasite.com/compiles.php

, regards