| Finished | ||||||
|---|---|---|---|---|---|---|
| Jhonno | Berserk | [Stockfish] dev | diff | N=5000 * | Generated 16666676/16666667 Games Elo: -20.25 +- 0.15 (95%) [N=16666667] Games: 16666676 W: 6340677 L: 7310748 D: 3015251 | |
| Rosent | Winter | qnet_382_d32 * | diff | 8.0+0.08 | LLR: 2.97 (-2.94, 2.94) [0.00, 5.00] Games: 17098 W: 4326 L: 4111 D: 8661 Ptnml(0-2): 375, 2021, 3560, 2200, 393 | Even larger quantized net. Might not yet be sufficiently mature. |
| Rosent | Winter | qnet_382_d24 * | diff | 40.0+0.40 | LLR: 2.95 (-2.94, 2.94) [0.00, 5.00] Games: 3404 W: 749 L: 622 D: 2033 Ptnml(0-2): 28, 331, 873, 426, 44 | Now that net is quantized, we try the larger size again. |
| Rosent | Winter | qnet_382_d24 * | diff | 8.0+0.08 | LLR: 2.96 (-2.94, 2.94) [0.00, 5.00] Games: 12870 W: 3262 L: 3070 D: 6538 Ptnml(0-2): 265, 1499, 2728, 1665, 278 | Now that net is quantized, we try the larger size again. |
| Rosent | Winter | quantize_net * | diff | 40.0+0.40 | LLR: 2.96 (-2.94, 2.94) [0.00, 5.00] Games: 1382 W: 324 L: 201 D: 857 Ptnml(0-2): 7, 121, 325, 218, 20 | Quantized network. Should be faster, at the cost of a little accuracy. |
| Rosent | Winter | quantize_net * | diff | 8.0+0.08 | LLR: 2.97 (-2.94, 2.94) [0.00, 5.00] Games: 1012 W: 301 L: 160 D: 551 Ptnml(0-2): 10, 80, 211, 169, 36 | Quantized network. Should be faster, at the cost of a little accuracy. |
| Vshche | Igel | static_tt_eval | diff | 5.0+0.05 | LLR: 2.96 (-2.94, 2.94) [0.00, 2.00] Games: 4524 W: 1273 L: 997 D: 2254 Ptnml(0-2): 12, 424, 1116, 696, 14 | 3.7.4: store static evaluation in TT |
| Vshche | Igel | static_tt_eval | diff | 10.0+0.10 | LLR: 2.95 (-2.94, 2.94) [0.00, 2.00] Games: 12652 W: 3161 L: 2871 D: 6620 Ptnml(0-2): 55, 1396, 3129, 1696, 50 | 3.7.4: store static evaluation in TT |
| Jhonno | Berserk | [Stockfish] dev | diff | N=7500 * | Generated 10000996/10000000 Games Elo: -9.47 +- 0.19 (95%) [N=10000000] Games: 10000996 W: 3924158 L: 4196621 D: 1880217 | |
| Vshche | Igel | draw_take2 | diff | 10.0+0.10 | LLR: 2.97 (-2.94, 2.94) [-3.00, 1.00] Games: 28970 W: 6894 L: 6803 D: 15273 Ptnml(0-2): 109, 3421, 7361, 3458, 136 | 3.7.3: yet another draw detection fix |
| Vshche | Igel | ep-take2 | diff | 10.0+0.10 | LLR: 2.99 (-2.94, 2.94) [-3.00, 1.00] Games: 140020 W: 33463 L: 33537 D: 73020 Ptnml(0-2): 557, 16389, 36185, 16329, 550 | 3.7.2: handle enpassants with tb |
| Jhonno | Berserk | [Stockfish] dev | diff | N=5000 * | Generated 1590/1000 Games Elo: -17.28 +- 15.57 (95%) [N=1000] Games: 1590 W: 615 L: 694 D: 281 | |
| Jhonno | Berserk | [Stockfish] dev | diff | N=7500 * | Generated 1214/1000 Games Elo: -6.30 +- 17.74 (95%) [N=1000] Games: 1214 W: 486 L: 508 D: 220 | |
| Jhonno | Berserk | [Stockfish] dev | diff | N=7500 * | Generated 1148/1000 Games Elo: -48.74 +- 18.54 (95%) [N=1000] Games: 1148 W: 392 L: 552 D: 204 | |
| Jhonno | Berserk | [Stockfish] dev | diff | N=7500 * | Generated 1206/1000 Games Elo: 18.74 +- 17.55 (95%) [N=1000] Games: 1206 W: 510 L: 445 D: 251 | |
| Jhonno | Berserk | [Stockfish] dev | diff | N=5000 | Generated 1470/1000 Games Elo: -64.06 +- 16.49 (95%) [N=1000] Games: 1470 W: 471 L: 739 D: 260 | |
| Vshche | Igel | tb-stat | diff | 10.0+0.10 | LLR: 3.01 (-2.94, 2.94) [-3.00, 1.00] Games: 9662 W: 2378 L: 2257 D: 5027 Ptnml(0-2): 41, 1083, 2452, 1224, 31 | 3.7.1: update statistics for tb moves |
| Vshche | Igel | iir-all-nodes | diff | 10.0+0.10 | LLR: 2.99 (-2.94, 2.94) [0.00, 2.00] Games: 41072 W: 10043 L: 9706 D: 21323 Ptnml(0-2): 175, 4825, 10228, 5104, 204 | 3.7.1: apply iid on all nodes |
| Vshche | Igel | draw_fix | diff | 60.0+0.60 | LLR: 2.97 (-2.94, 2.94) [-3.00, 1.00] Games: 32586 W: 5991 L: 5939 D: 20656 Ptnml(0-2): 1, 1601, 13037, 1653, 1 | 3.6.39: fix draw detection |
| Jhonno | Berserk | refactor/remove-setjmp-search | diff | 10.0+0.10 | LLR: 2.94 (-2.25, 2.89) [-2.00, 0.00] Games: 31618 W: 7773 L: 7571 D: 16274 Ptnml(0-2): 76, 3287, 8860, 3531, 55 | refactor: remove setjmp from search |
| Jhonno | Berserk | speedups | diff | 10.0+0.10 | LLR: 2.94 (-2.25, 2.89) [0.00, 2.00] Games: 11046 W: 2816 L: 2549 D: 5681 Ptnml(0-2): 28, 1086, 3028, 1353, 28 | perf: drop the pext slider path and the -pext binaries Since AVX2 builds compute slider attacks by hyperbola quintessence, the pext lookup only ever applied to targets that have BMI2 but not AVX2, which is not a combination worth carrying a second slider implementation for. Stockfish reached the same point and removed its pext path outright. That also collapses the release matrix. avx2-pext differed from avx2 only by -mbmi2, and avx512-pext existed because every AVX-512 part has BMI2, so both names now describe nothing. The Windows build script ships four binaries instead of five: x86-64, sse41, avx2, avx512. No functional change on any AVX2 host: the emitted code for a native build is byte identical before and after. Co-authored-by: Claude Agent <claude@home.arpa> |
| Vshche | Igel | draw_fix | diff | 10.0+0.10 | LLR: 2.97 (-2.94, 2.94) [-3.00, 1.00] Games: 11210 W: 2172 L: 2087 D: 6951 Ptnml(0-2): 2, 742, 4035, 821, 5 | 3.6.39: fix draw detection |
| Vshche | Igel | history-follow-decay-operand | diff | 60.0+0.60 | LLR: 3.06 (-2.94, 2.94) [-3.00, 1.00] Games: 18668 W: 4402 L: 4297 D: 9969 Ptnml(0-2): 38, 2099, 4947, 2220, 30 | 3.6.38: fix history overflow |
| Rosent | Winter | net347_d24f64 * | diff | N=40000 | Elo: 25.02 +- 7.56 (95%) [N=4000] Games: 4020 W: 1277 L: 988 D: 1755 Ptnml(0-2): 95, 414, 767, 575, 159 | Preliminary fixed node test with a potential larger net. |
| Rosent | Winter | net347_W50 * | diff | 40.0+0.40 | LLR: 2.96 (-2.94, 2.94) [0.00, 5.00] Games: 4838 W: 986 L: 852 D: 3000 Ptnml(0-2): 44, 487, 1244, 579, 65 | Trained with more data. |