Tests

New test

Only enabled users can create tests

Finished 25 shown

Author	Engine	Identifier	Code	Time control	Results
nocturn9x	Heimdall	rep-collision-fix #9123 Passed Error	diff ↗	8.0+0.08	LLR: 2.90 (-2.25, 2.89) [-5.00, 0.00] Games: 9330 W: 2325 L: 2257 D: 4748 Ptnml(0-2): 56, 955, 2550, 1073, 31
Yeet silly debug easter egg to use faster autovec repetition comparisons + clang LTO black magic (bench 5611448)
nocturn9x	Heimdall	rep-collision-fix #9119 Finished	diff ↗	8.0+0.08	LLR: 0.97 (-2.25, 2.89) [-5.00, 0.00] Games: 4930 W: 1232 L: 1216 D: 2482 Ptnml(0-2): 23, 535, 1327, 563, 17
Double check zobrist collisions using our position stack (bench 5611448) Co-authored-by: GPT 6 <codex@openai.com> Not actually coauthored by ze slopper btw, I just had it fix missing benches and it upid. Keeping this up for accountability purposes
MasterStruct	Eunoia	feat/aw-centering #9118 Finished	diff ↗	40.0+0.40	LLR: 1.46 (-2.25, 2.89) [0.00, 5.00] Games: 1416 W: 429 L: 356 D: 631 Ptnml(0-2): 26, 146, 310, 181, 45
Widen aspiration windows based on score
MasterStruct	Eunoia	feat/killer-heuristic #9114 Finished	diff ↗	8.0+0.08	LLR: -0.23 (-2.25, 2.89) [0.00, 5.00] Games: 17262 W: 5028 L: 4953 D: 7281 Ptnml(0-2): 512, 2018, 3513, 2059, 529
Use NodeStack for killer moves
nocturn9x	Heimdall	worker-pv-score-fix #7937 Passed	diff ↗	8.0+0.08	LLR: 2.91 (-2.25, 2.89) [-5.00, 0.00] Games: 6392 W: 1600 L: 1520 D: 3272 Ptnml(0-2): 15, 733, 1624, 805, 19
Use completed PV scores when selecting workers (bench 5551733) Compare workers using scores saved with their last completed primary PV, and report the selected PV's saved score. Live root scores may belong to secondary variations or interrupted searches. Ignore workers without a completed primary PV. Co-authored-by: Codex (GPT-6 Astra) <codex@openai.com>
nocturn9x	Heimdall	tyrfing-12ksn #9048 Passed Error	diff ↗	40.0+0.40	LLR: 2.90 (-2.25, 2.89) [0.00, 3.00] Games: 12500 W: 3242 L: 3040 D: 6218 Ptnml(0-2): 32, 1440, 3111, 1628, 39
Merge master and select A8607157 (bench 5611448) Bring tyrfing-12ksn up to date with master and select the new A8607157 network with the unchanged multilayer TI architecture. Keep the weights local in networks/files/tyrfing-12ksn-v2.bin; network publication remains deferred until the net is merged. Calibrate on all 7,153,653 positions in lichess-big3-resolved.book. At scale 305, getStats reports absolute means of 927.90 for master's 510EEBF8 network and 1004.55 for A8607157. getScale recommends 281.727639, rounded to 282. At scale 282, the new absolute mean is 928.76. Update the documented default scale. Validation: make dev and the runtime-check build pass. Scalar and AVX2 NNUE tests each pass 17,731 incremental/full-refresh comparisons with identical checksums. Python regressions pass (99 tests, 5 skipped) with Git signing disabled for temporary test commits. The repository commit hook also passes make test. The optimized depth-13 bench is 5611448 nodes. Network SHA-256: a8607157fbb8ac900ac7fe25d45d772509db9ecf3d9b5cb5000e5a581e1d0890 Co-authored-by: GPT 6 <codex@openai.com>
MasterStruct	Eunoia	feat/aw-centering #9115 Passed	diff ↗	8.0+0.08	LLR: 2.97 (-2.25, 2.89) [0.00, 5.00] Games: 5564 W: 1692 L: 1530 D: 2342 Ptnml(0-2): 130, 565, 1264, 659, 164
Widen aspiration windows based on score
MasterStruct	Eunoia	feat/inc-fp-depth #9116 Passed	diff ↗	8.0+0.08	LLR: 3.00 (-2.25, 2.89) [0.00, 5.00] Games: 5516 W: 1708 L: 1536 D: 2272 Ptnml(0-2): 142, 621, 1110, 693, 192
Increase Futility Pruning max depth
MasterStruct	Eunoia	feat/nmp-margin #9117 Finished	diff ↗	8.0+0.08	LLR: -1.46 (-2.25, 2.89) [0.00, 5.00] Games: 1312 W: 348 L: 413 D: 551 Ptnml(0-2): 52, 173, 252, 146, 33
Increase NMP margin
MasterStruct	Eunoia	feat/lmr-improving #9113 Passed	diff ↗	8.0+0.08	LLR: 2.90 (-2.25, 2.89) [0.00, 5.00] Games: 7582 W: 2260 L: 2083 D: 3239 Ptnml(0-2): 203, 871, 1504, 972, 241
Reduce more in LMR when not improving
MasterStruct	Eunoia	feat/lmp-improving #9112 Passed	diff ↗	8.0+0.08	LLR: 2.90 (-2.25, 2.89) [0.00, 5.00] Games: 5698 W: 1755 L: 1583 D: 2360 Ptnml(0-2): 155, 672, 1082, 726, 214
Add LMP improving
dan	Hobbes	main #7901 Passed Error	diff ↗	N=12000	Generated 5000056/5000000 Games Elo: 0.03 +- 0.24 (95%) [N=5000000] Games: 5000056 W: 1606277 L: 1605903 D: 1787876
MasterStruct	Eunoia	feat/rfp-improving #9111 Passed	diff ↗	8.0+0.08	LLR: 2.94 (-2.25, 2.89) [0.00, 5.00] Games: 1894 W: 662 L: 502 D: 730 Ptnml(0-2): 56, 167, 376, 257, 91
Add RFP improving
MasterStruct	Eunoia	feat/killer-heuristic #9110 Finished	diff ↗	8.0+0.08	LLR: 0.87 (-2.25, 2.89) [0.00, 5.00] Games: 9774 W: 2948 L: 2856 D: 3970 Ptnml(0-2): 316, 1101, 1957, 1201, 312
Skip history pruning for killers
dan	Hobbes	feature/integral-tt-cut-agree #9109 Failed	diff ↗	8.0+0.08	LLR: -2.30 (-2.25, 2.89) [0.00, 3.00] Games: 12768 W: 3009 L: 3120 D: 6639 Ptnml(0-2): 68, 1544, 3278, 1419, 75
dan	Hobbes	feature/re-corr-moveloop #9108 Failed	diff ↗	8.0+0.08	LLR: -2.25 (-2.25, 2.89) [0.00, 3.00] Games: 9688 W: 2238 L: 2349 D: 5101 Ptnml(0-2): 27, 1195, 2506, 1094, 22
Ellie	WakWak	rumali #9107 Finished Duck960	diff ↗	N=2500000	LLR: -0.15 (-2.25, 2.89) [0.00, 4.20] Games: 56 W: 19 L: 24 D: 13 Ptnml(0-2): 1, 5, 20, 2, 0
2048 l1
Ellie	WakWak	rumali #9106 Failed Duck960	diff ↗	N=25000	LLR: -2.39 (-2.25, 2.89) [0.00, 4.20] Games: 2876 W: 969 L: 1103 D: 804 Ptnml(0-2): 156, 287, 638, 249, 108
2048 l1
MasterStruct	Eunoia	feat/razoring #9104 Finished	diff ↗	8.0+0.08	LLR: -0.96 (-2.25, 2.89) [0.00, 5.00] Games: 12832 W: 3743 L: 3725 D: 5364 Ptnml(0-2): 407, 1521, 2539, 1545, 404
Fix razoring in check
MasterStruct	Eunoia	feat/killer-heuristic #9105 Failed	diff ↗	8.0+0.08	LLR: -2.35 (-2.25, 2.89) [0.00, 5.00] Games: 912 W: 226 L: 341 D: 345 Ptnml(0-2): 49, 135, 173, 80, 19
Clear killer table before searching
Tecci	WakWak	1390042C22A3CAFB #9103 Passed Duck960	diff ↗	8.0+0.08	LLR: 2.90 (-2.25, 2.89) [-4.20, 0.00] Games: 9138 W: 2952 L: 2845 D: 3341 Ptnml(0-2): 154, 844, 2493, 897, 181
strength option (#235)
Tecci	WakWak	silly-see-idea #9102 Finished Duck960	diff ↗	8.0+0.08	LLR: -1.20 (-2.25, 2.89) [0.00, 4.20] Games: 882 W: 271 L: 331 D: 280 Ptnml(0-2): 31, 98, 229, 66, 17
don't consider the previous duck move in SEE
Ellie	WakWak	1390042C22A3CAFB #9101 Failed Duck960	diff ↗	8.0+0.08	LLR: -2.27 (-2.25, 2.89) [-4.20, 0.00] Games: 10474 W: 3225 L: 3372 D: 3877 Ptnml(0-2): 199, 1092, 2787, 975, 184
strength option (#235)
MasterStruct	Eunoia	fix/zero-searched #9098 Passed	diff ↗	8.0+0.08	LLR: 2.93 (-2.25, 2.89) [-10.00, 0.00] Games: 2254 W: 665 L: 621 D: 968 Ptnml(0-2): 45, 240, 525, 260, 57
Return alpha if legal moves exist but none searched
Ellie	WakWak	search-exclude-moves #9100 Finished Duck960	diff ↗	8.0+0.08	LLR: 1.07 (-2.25, 2.89) [-4.20, 0.00] Games: 4266 W: 1387 L: 1349 D: 1530 Ptnml(0-2): 83, 424, 1096, 432, 98
ok i think i fixed the fup