SPCC: Longtime testrun of Fat Titz 2 vs. SF finished

[pohl4711]

Experimental testrun (1000 games!) of Fat Titz 2 vs. Stockfish 220113 with very long thinking-time finished. Goal: Find out, if Fat Titz 2 can benefit from his bigger nnue-net (compared to Stockfish), when the thinking-time is very long... Thinking-time: 20min+10sec, average game-duration 65 minutes!!!


https://www.sp-cc.de

(Perhaps you have to clear your browsercache or reload the website)

[Uri Blass]

The thinking time 20+10 is not very long considering the fact you use no ponder and only one core.
No problem with testing but I am against the definition of very long time control.

I found only one game when stockrfish won with black(other wins of both sides are win of white and no problem with it and I understand the opening are unbalanced).

It may be interesting if somebody use the same initial positions with longer time control to compare.

The consequence that Fat Titz 2 does not benefit from his much bigger nnue-net (compared to Stockfish) is also not correct conclusion and we can only say that 1000 games at 1200+10 games are not enough to know.

Maybe the difference at 1200+10 is too small to measure with 1000 games and we need more cores or longer time control to measure a difference.

[pgn][Event "Fat Titz 2 vs SF longtime"]
[Site "?"]
[Date "2022.01.15"]
[Round "8"]
[White "Fat Titz 2 bmi2"]
[Black "Stockfish 220113 bmi2"]
[Result "0-1"]
[ECO "B00"]
[PlyCount "215"]
[EventDate "2022.??.??"]
[TimeControl "1200+10"]

1. e4 {book} d6 {book} 2. d4 {book} g6 {book} 3. c4 {book} Bg7 {book} 4. Nc3 {
book} Nd7 {book} 5. f4 {book} c5 {book} 6. d5 {book} Bxc3+ {book} 7. bxc3 {book
} Ngf6 {book} 8. Bd3 {book} Qa5 {book} 9. Bd2 {[%eval 187,32] 93s} b5 {[%eval
-129,31] 73s} 10. Nf3 {[%eval 200,30] [%emt 0:00:28]} bxc4 {[%eval -115,29]
[%emt 0:00:22]} 11. Bc2 {[%eval 191,29] [%emt 0:00:26]} Rb8 {[%eval -117,29]
[%emt 0:00:26]} 12. O-O {[%eval 179,30] [%emt 0:00:34]} Rb6 {[%eval -138,30]
[%emt 0:00:38]} 13. a4 {[%eval 166,26] [%emt 0:00:45]} Rf8 {[%eval -122,32] 77s
} 14. Be1 {[%eval 167,29] 91s} Bb7 {[%eval -99,29] [%emt 0:00:55]} 15. Nd2 {
[%eval 187,27] [%emt 0:00:20]} Ba6 {[%eval -93,30] [%emt 0:00:20]} 16. Nf3 {
[%eval 156,31] 139s} Bb7 {[%eval -100,31] [%emt 0:00:50]} 17. Nd2 {[%eval 171,
29] [%emt 0:00:33]} Ba6 {[%eval -106,31] [%emt 0:00:25]} 18. Nb1 {[%eval 184,
29] [%emt 0:00:24]} Kd8 {[%eval -99,27] [%emt 0:00:26]} 19. Na3 {[%eval 182,26]
[%emt 0:00:25]} Ne8 {[%eval -93,27] [%emt 0:00:25]} 20. h3 {[%eval 169,27]
[%emt 0:00:41]} f6 {[%eval -87,27] [%emt 0:00:31]} 21. Kh2 {[%eval 170,26]
[%emt 0:00:22]} Nc7 {[%eval -87,32] 81s} 22. Nb5 {[%eval 167,28] [%emt 0:00:27]
} Rb7 {[%eval -112,30] [%emt 0:00:59]} 23. Bd2 {[%eval 163,26] [%emt 0:00:22]}
Qb6 {[%eval -87,29] [%emt 0:00:24]} 24. Na3 {[%eval 167,27] [%emt 0:00:22]} Qa5
{[%eval -90,28] [%emt 0:00:26]} 25. Be1 {[%eval 162,29] [%emt 0:00:22]} Rb8 {
[%eval -88,31] 85s} 26. Rf2 {[%eval 162,31] 127s} Rb7 {[%eval -102,30] [%emt 0:
00:32]} 27. Ra2 {[%eval 163,24] [%emt 0:00:13]} Rb6 {[%eval -110,29] 73s} 28.
Rf3 {[%eval 174,30] 83s} Rb8 {[%eval -120,29] [%emt 0:00:41]} 29. Re3 {[%eval
172,28] 63s} Rf7 {[%eval -129,28] [%emt 0:00:58]} 30. Rg3 {[%eval 195,23]
[%emt 0:00:14]} Qb6 {[%eval -132,29] [%emt 0:00:15]} 31. a5 {[%eval 227,27] 74s
} Qxa5 {[%eval -129,24] [%emt 0:00:29]} 32. Re3 {[%eval 225,27] [%emt 0:00:33]}
Ne8 {[%eval -136,30] [%emt 0:00:48]} 33. Bg3 {[%eval 222,24] [%emt 0:00:18]}
Qc7 {[%eval -144,25] [%emt 0:00:17]} 34. Ba4 {[%eval 220,25] [%emt 0:00:27]}
Ng7 {[%eval -157,30] [%emt 0:00:51]} 35. Kg1 {[%eval 202,27] [%emt 0:00:58]}
Nb6 {[%eval -140,28] [%emt 0:00:18]} 36. Bc6 {[%eval 219,22] [%emt 0:00:12]}
Nc8 {[%eval -138,27] [%emt 0:00:17]} 37. Re1 {[%eval 211,23] [%emt 0:00:20]}
Rb6 {[%eval -139,29] [%emt 0:00:57]} 38. Ba4 {[%eval 197,27] [%emt 0:00:49]} f5
{[%eval -116,28] [%emt 0:00:36]} 39. e5 {[%eval 203,29] [%emt 0:00:17]} Nh5 {
[%eval -125,29] [%emt 0:00:17]} 40. Bh2 {[%eval 200,25] [%emt 0:00:17]} Qb7 {
[%eval -105,25] [%emt 0:00:18]} 41. Rd2 {[%eval 187,28] [%emt 0:00:17]} Kc7 {
[%eval -130,31] [%emt 0:00:53]} 42. Qc2 {[%eval 200,28] [%emt 0:00:16]} Rf8 {
[%eval -139,28] [%emt 0:00:16]} 43. Qa2 {[%eval 199,28] [%emt 0:00:41]} Rd8 {
[%eval -105,28] [%emt 0:00:24]} 44. Nxc4 {[%eval 193,26] [%emt 0:00:32]} Bxc4 {
[%eval -138,31] [%emt 0:00:44]} 45. Qxc4 {[%eval 153,28] [%emt 0:00:34]} Qa6 {
[%eval -130,30] [%emt 0:00:19]} 46. Qa2 {[%eval 161,25] [%emt 0:00:17]} Qa5 {
[%eval -132,32] [%emt 0:00:23]} 47. Qc4 {[%eval 162,29] [%emt 0:00:47]} Qa6 {
[%eval -127,32] [%emt 0:00:24]} 48. Qa2 {[%eval 190,25] 3.6s} Qa5 {[%eval -136,
33] [%emt 0:00:24]} 49. Qc2 {[%eval 207,26] 5.7s} Qa6 {[%eval -150,32] [%emt 0:
00:22]} 50. Bc6 {[%eval 175,28] [%emt 0:00:49]} Qc4 {[%eval -128,29] [%emt 0:
00:15]} 51. Ra1 {[%eval 163,23] 7.1s} Rf8 {[%eval -110,25] [%emt 0:00:10]} 52.
Qc1 {[%eval 189,22] 3.4s} a6 {[%eval -89,24] 9.8s} 53. Ra4 {[%eval 63,23]
[%emt 0:00:28]} Qb3 {[%eval -45,25] [%emt 0:00:13]} 54. Raa2 {[%eval 109,25] 6.
2s} Qc4 {[%eval -34,25] [%emt 0:00:18]} 55. Rdb2 {[%eval 81,24] [%emt 0:00:17]}
dxe5 {[%eval 0,30] [%emt 0:00:10]} 56. fxe5 {[%eval 0,19] 3.0s} f4 {[%eval 0,
32] [%emt 0:00:17]} 57. Ra1 {[%eval 0,22] 7.4s} Rxb2 {[%eval 0,27] 8.5s} 58.
Qxb2 {[%eval 0,21] 1.8s} Nb6 {[%eval 0,28] 8.6s} 59. Rd1 {[%eval -2,23] 9.4s}
Rb8 {[%eval 51,25] 8.2s} 60. Qa3 {[%eval 21,23] 7.7s} g5 {[%eval 131,25] 9.0s}
61. Qa5 {[%eval 0,23] 5.4s} g4 {[%eval 368,25] 9.6s} 62. hxg4 {[%eval -294,22]
[%emt 0:00:12]} Qe2 {[%eval 346,28] 9.7s} 63. Ba4 {[%eval -295,24] 4.6s} Qe3+ {
[%eval 389,28] 9.3s} 64. Kh1 {[%eval -406,27] [%emt 0:00:13]} Ng3+ {[%eval 402,
26] 7.6s} 65. Bxg3 {[%eval -436,23] 1.1s} fxg3 {[%eval 428,28] 8.7s} 66. g5 {
[%eval -437,28] 9.7s} Qxg5 {[%eval 453,30] 8.8s} 67. Qxc5+ {[%eval -469,32]
[%emt 0:00:17]} Kb7 {[%eval 455,30] 9.1s} 68. Kg1 {[%eval -470,26] 8.2s} Qh4 {
[%eval 475,28] [%emt 0:00:11]} 69. Bc6+ {[%eval -472,28] 4.7s} Ka7 {[%eval 469,
25] 8.4s} 70. Qb4 {[%eval -483,30] 8.1s} Qh2+ {[%eval 493,26] [%emt 0:00:10]}
71. Kf1 {[%eval -486,24] 0.82s} Qh1+ {[%eval 484,25] 8.8s} 72. Ke2 {[%eval
-486,24] 0.56s} Qxg2+ {[%eval 507,24] 8.8s} 73. Kd3 {[%eval -480,31] [%emt 0:
00:12]} Qf3+ {[%eval 512,26] [%emt 0:00:14]} 74. Kd2 {[%eval -519,30] [%emt 0:
00:19]} Qf2+ {[%eval 525,25] 9.0s} 75. Kd3 {[%eval -545,29] [%emt 0:00:17]}
Qf3+ {[%eval 523,28] [%emt 0:00:23]} 76. Kd2 {[%eval -559,30] [%emt 0:00:32]}
Qf2+ {[%eval 507,25] 6.7s} 77. Kd3 {[%eval -580,28] 5.8s} Nxd5 {[%eval 513,28]
[%emt 0:00:13]} 78. Qd4+ {[%eval -600,26] 6.9s} Qxd4+ {[%eval 528,26] 7.3s} 79.
Kxd4 {[%eval -596,26] 3.4s} Nc7 {[%eval 529,28] 9.9s} 80. Rh1 {[%eval -318,24]
[%emt 0:00:24]} a5 {[%eval 527,29] 8.2s} 81. Kc4 {[%eval -405,22] [%emt 0:00:
15]} Rg8 {[%eval 528,30] 8.6s} 82. Rxh7 {[%eval -421,24] 3.2s} g2 {[%eval 563,
29] [%emt 0:00:15]} 83. Bxg2 {[%eval -419,25] 5.5s} Rxg2 {[%eval 563,27] 10.0s}
84. Rxe7 {[%eval -417,29] 4.2s} Kb6 {[%eval 584,27] [%emt 0:00:11]} 85. Rh7 {
[%eval -445,29] [%emt 0:00:20]} Rg4+ {[%eval 574,25] [%emt 0:00:14]} 86. Kd3 {
[%eval -448,29] [%emt 0:00:15]} Rg5 {[%eval 609,27] 7.5s} 87. Kc4 {[%eval -464,
28] 4.8s} Rxe5 {[%eval 633,24] 7.1s} 88. Kb3 {[%eval -535,26] [%emt 0:00:14]}
Nd5 {[%eval 667,26] 9.0s} 89. Rh6+ {[%eval -535,27] [%emt 0:00:22]} Kb7 {
[%eval 741,22] 7.9s} 90. Kc4 {[%eval -581,24] [%emt 0:00:12]} Nb6+ {[%eval 821,
22] 9.8s} 91. Kb3 {[%eval -589,20] 1.0s} Ka6 {[%eval 896,24] 9.2s} 92. Kc2 {
[%eval -700,23] [%emt 0:00:18]} Kb5 {[%eval 936,23] 8.8s} 93. Rh8 {[%eval -761,
22] [%emt 0:00:11]} a4 {[%eval 990,23] 9.2s} 94. Rb8 {[%eval -783,20] 3.5s}
Re2+ {[%eval 1047,22] [%emt 0:00:13]} 95. Kb1 {[%eval -897,23] 6.9s} Kc5 {
[%eval 5508,31] [%emt 0:00:18]} 96. Kc1 {[%eval -1014,24] 6.4s} a3 {[%eval
32678,35] 8.3s} 97. Kb1 {[%eval -921,24] 7.2s} Na4 {[%eval 32702,33] [%emt 0:
00:19]} 98. Rc8+ {[%eval -32707,51] [%emt 0:00:17]} Kb5 {[%eval 32710,35]
[%emt 0:00:14]} 99. Rb8+ {[%eval -32711,41] 2.5s} Kc4 {[%eval 32714,34] [%emt
0:00:14]} 100. Rc8+ {[%eval -32715,44] 6.4s} Nc5 {[%eval 32718,33] [%emt 0:00:
17]} 101. Kc1 {[%eval -32719,44] 3.3s} a2 {[%eval 32722,37] [%emt 0:00:15]}
102. Ra8 {[%eval -32723,43] [%emt 0:00:16]} Nb3+ {[%eval 32726,37] [%emt 0:00:
14]} 103. Kd1 {[%eval -32727,43] 0.56s} Rb2 {[%eval 32730,37] [%emt 0:00:16]}
104. Rc8+ {[%eval -32731,50] 4.5s} Kd3 {[%eval 32734,31] [%emt 0:00:17]} 105.
Rd8+ {[%eval -32735,49] [%emt 0:00:18]} Kxc3 {[%eval 32738,38] [%emt 0:00:16]}
106. Ra8 {[%eval -32739,46] [%emt 0:00:17]} Kd3 {[%eval 32742,40] [%emt 0:00:
15]} 107. Ke1 {[%eval -32743,70] 3.1s} a1=Q+ {[%eval 32746,44] [%emt 0:00:14]}
108. Rxa1 {adjudication} 0-1
[/pgn]

[dkappe]

Code: Select all

     Program                 Elo    +    -   Games   Score  Draws
   1 Fat Titz 2 bmi2       : 3802   7    7   1000    50.3%  51.7%
   2 Stockfish 220113 bmi2 : 3800   7    7   1000    49.8%  51.7%

Fairly even. Not bad for a Stockfish clone with a bigger net.

[Jouni]

Thanks for testing! I don't get why Sopel wastes time with this old joke still?

[gerhard]

forum3/viewtopic.php?f=2&t=79112&start=20#p918915

.... dont waste your time

[dkappe]

Thanks for testing! I don't get why Sopel wastes time with this old joke still?

Leaving the joke aside (which is a bit of a problem for the Stockfish brand), training a much bigger stockfish NNUE net is an interesting challenge.