OpenBench

OpenBench Testing Framework

Active : 10 Machines / 708 Threads / 371.86 MNPS
Priority 0
World	Reckless	fds-reduced-depth	diff	8.0+0.08	LLR: -0.31 (-2.25, 2.89) [-2.75, 0.25] Games: 11508 W: 2885 L: 2924 D: 5699 Ptnml(0-2): 29, 1373, 2991, 1330, 31
Sapher	Reckless	conti-scale	diff	8.0+0.08	LLR: -1.09 (-2.25, 2.89) [0.00, 3.00] Games: 64466 W: 16236 L: 16159 D: 32071 Ptnml(0-2): 170, 7536, 16728, 7645, 154	Take 1 ~6000 iterations SPSA
World	Reckless	lmr-reduced-depth	diff	8.0+0.08	LLR: -1.92 (-2.25, 2.89) [0.00, 3.00] Games: 10838 W: 2626 L: 2715 D: 5497 Ptnml(0-2): 29, 1321, 2803, 1242, 24
Priority -1
World	Reckless	saphereye-1	diff	4.0+0.04	LLR: -1.35 (-2.25, 2.89) [-2.75, 0.00] Games: 36346 W: 9274 L: 9433 D: 17639 Ptnml(0-2): 162, 4087, 9836, 3924, 164
Priority -5
Sapher	Reckless	conti-scale	diff	40.0+0.40	Tuning 9 Parameters 6767/25000 Iterations 108276/400000 Games Played	Take 1.1

Finished
Proton	Reckless	ps_eplegal101	diff	8.0+0.08	LLR: -0.41 (-2.25, 2.89) [-2.75, 0.25] Games: 1540 W: 374 L: 399 D: 767 Ptnml(0-2): 1, 176, 443, 147, 3	Merge branch 'main' into ps_eplegal101
Proton	Reckless	ps_eplegal100	diff	8.0+0.08	LLR: -0.21 (-2.25, 2.89) [-2.75, 0.25] Games: 37814 W: 9496 L: 9578 D: 18740 Ptnml(0-2): 110, 4527, 9714, 4447, 109	simplify ep move generation legality.
Sapher	Reckless	rfp-cleaning	diff	8.0+0.08	LLR: -0.79 (-2.25, 2.89) [0.00, 3.00] Games: 14624 W: 3712 L: 3726 D: 7186 Ptnml(0-2): 42, 1609, 4015, 1613, 33	Take 1
87	Reckless	triplet-tune	diff	40.0+0.40	Tuning 10 Parameters 25000/25000 Iterations 100002/100000 Games Played
Joshka	Reckless	joshka/load-nnue-parameters-with-zerocopy	diff	4.0+0.04	LLR: -2.29 (-2.25, 2.89) [0.00, 3.00] Games: 14098 W: 3597 L: 3702 D: 6799 Ptnml(0-2): 67, 1658, 3701, 1559, 64	Load NNUE parameters with zerocopy Use zerocopy for the static NNUE model layout instead of an explicit transmute
Proton	Reckless	ps_capturesq102	diff	8.0+0.08	LLR: 3.05 (-2.25, 2.89) [-2.75, 0.25] Games: 37104 W: 9455 L: 9348 D: 18301 Ptnml(0-2): 83, 4001, 10285, 4092, 91	Cleanup using capture_sq.
Proton	Reckless	ps_castling103	diff	8.0+0.08	LLR: 3.01 (-2.25, 2.89) [-2.75, 0.25] Games: 16150 W: 4126 L: 3979 D: 8045 Ptnml(0-2): 52, 1773, 4285, 1906, 59	More simple.
Cj5716	Reckless	more-permissive-singext	diff	40.0+0.40	LLR: -2.36 (-2.25, 2.89) [0.00, 3.00] Games: 41674 W: 10304 L: 10349 D: 21021 Ptnml(0-2): 17, 4764, 11316, 4727, 13
Cj5716	Reckless	use-future-adjust-past	diff	8.0+0.08	LLR: -2.25 (-2.25, 2.89) [0.00, 3.00] Games: 23548 W: 5904 L: 5985 D: 11659 Ptnml(0-2): 65, 2824, 6073, 2751, 61	Reduce bonus if we were "carried" by the synergy between our children and the evaluated move
Cj5716	Reckless	more-permissive-singext	diff	8.0+0.08	LLR: 2.90 (-2.25, 2.89) [0.00, 3.00] Games: 10022 W: 2627 L: 2433 D: 4962 Ptnml(0-2): 27, 1105, 2554, 1297, 28
Proton	Reckless	ps_nonpawns102	diff	40.0+0.40	LLR: 2.95 (-2.25, 2.89) [-2.75, 0.25] Games: 52454 W: 13118 L: 13047 D: 26289 Ptnml(0-2): 30, 5476, 15140, 5555, 26	600
Peregr	Reckless	fix-tt-again	diff	5.0+0.05	LLR: -2.30 (-2.25, 2.89) [0.00, 2.50] Games: 22672 W: 5548 L: 5663 D: 11461 Ptnml(0-2): 23, 2621, 6167, 2498, 27	Try fixing overlap between empty and Some in TT again
Peregr	Reckless	fix-tt-again	diff	8.0+0.08	LLR: -0.77 (-2.25, 2.89) [-2.75, 0.25] Games: 49054 W: 12272 L: 12408 D: 24374 Ptnml(0-2): 140, 5896, 12613, 5716, 162	Try fixing overlap between empty and Some in TT again
87	Reckless	rm-minor-corrhist	diff	40.0+0.40	LLR: 2.95 (-2.25, 2.89) [-2.75, 0.25] Games: 87334 W: 21706 L: 21693 D: 43935 Ptnml(0-2): 37, 9783, 24004, 9816, 27
Proton	Reckless	ps_nonpawns104	diff	8.0+0.08	LLR: -2.48 (-2.25, 2.89) [0.00, 3.00] Games: 40778 W: 10258 L: 10312 D: 20208 Ptnml(0-2): 114, 4432, 11342, 4396, 105	remove non_pawns, but verify later.
87	Reckless	rm-minor-corrhist	diff	8.0+0.08	LLR: 2.93 (-2.25, 2.89) [-2.75, 0.25] Games: 20930 W: 5274 L: 5139 D: 10517 Ptnml(0-2): 54, 2432, 5368, 2547, 64
87	Reckless	triplet4	diff	8.0+0.08	LLR: -1.42 (-2.25, 2.89) [0.00, 3.00] Games: 11908 W: 2999 L: 3057 D: 5852 Ptnml(0-2): 36, 1451, 3043, 1383, 41
Peregr	Reckless	qs-explosive	diff	8.0+0.08	LLR: -2.27 (-2.25, 2.89) [0.00, 3.00] Games: 45566 W: 11557 L: 11591 D: 22418 Ptnml(0-2): 154, 5416, 11669, 5398, 146
Peregr	Reckless	depth-one-lower	diff	8.0+0.08	LLR: -2.30 (-2.25, 2.89) [0.00, 3.00] Games: 16392 W: 4096 L: 4195 D: 8101 Ptnml(0-2): 50, 1962, 4270, 1865, 49
Peregr	Reckless	benefit-from-quiets	diff	8.0+0.08	LLR: -2.29 (-2.25, 2.89) [0.00, 3.00] Games: 50600 W: 12687 L: 12711 D: 25202 Ptnml(0-2): 149, 6086, 12871, 6028, 166	make use of the quiets tried in qsearch to update corrhist as if they were depth 1 searches
Proton	Reckless	ps_nonpawns103	diff	8.0+0.08	LLR: 0.18 (-2.25, 2.89) [-2.75, 0.25] Games: 8550 W: 2177 L: 2182 D: 4191 Ptnml(0-2): 27, 1007, 2213, 1000, 28	just remove non-pawns.
Peregr	Reckless	bad-noisy-pawn	diff	8.0+0.08	LLR: -0.67 (-2.25, 2.89) [0.00, 3.00] Games: 1964 W: 488 L: 523 D: 953 Ptnml(0-2): 6, 247, 508, 218, 3	Assume that in qsearch move ordering, winning a pawn is never bad
Joshka	Reckless	joshka/split-worker-state-from-search-state	diff	4.0+0.04	LLR: 0.00 (-2.25, 2.89) [0.00, 3.00] Games: 0 W: 0 L: 0 D: 0 Ptnml(0-2): 0, 0, 0, 0, 0	Replaces a small set of layout casts with `zerocopy`, and refactors the threadpool so helper threads own their state and communicate through explicit request/result types instead of shared `ThreadData` access. Bench: `3142485`.
Proton	Reckless	ps_nonpawns102	diff	40.0+0.40	LLR: 2.92 (-2.25, 2.89) [-2.75, 0.25] Games: 9090 W: 2296 L: 2154 D: 4640 Ptnml(0-2): 4, 889, 2620, 1025, 7	600
Proton	Reckless	ps_nonpawns102	diff	8.0+0.08	LLR: 2.92 (-2.25, 2.89) [-2.75, 0.25] Games: 23032 W: 5919 L: 5796 D: 11317 Ptnml(0-2): 40, 2446, 6420, 2571, 39	600

1 2 3 395 396 397