Stockfish Testing Queue

Finished - 4981 tests

17-04-28 sg passed_pawn_break2 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 16752 W: 3141 L: 2944 D: 10667
sprt @ 10+0.1 th 1 Version of my last try with half eval (+ fix of assert in evaluate_passed_pawns). Set prio to -1 to wait for the other version.
17-04-27 sg passed_pawn_break diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 5934 W: 1160 L: 1009 D: 3765
sprt @ 10+0.1 th 1 Bonus for pawn sacrifice which creates an advanced passed pawn
17-04-23 sg sort_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 21408 W: 3833 L: 3917 D: 13658
sprt @ 10+0.1 th 1 My LTC test seems not good so far so perhaps more sorting is needed at higher depths. Test against passed sortLinearRebase patch.
17-04-23 sg sort_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 90756 W: 16578 L: 16424 D: 57754
sprt @ 10+0.1 th 1 Last try had a break even point at depth 5 which seems to low. With weight 4500 break even is at depth 9 (depth >9 more sorting, depth <9 less sorting).
17-04-24 sg sort_tweak3 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 11829 W: 2089 L: 2205 D: 7535
sprt @ 10+0.1 th 1 Ok make a last try. Even less sorting for all depths with weight 3000
17-04-23 sg sort_tweak diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 58576 W: 7664 L: 7658 D: 43254
sprt @ 60+0.6 th 1 LTC: Less sorting: shift by 4000. So we have for depth 1 the limit zero (the same as in current master). Test against passed sortLinearRebase patch.
17-04-20 sg master diff
ELO: 15.17 +-1.6 (95%) LOS: 100.0%
Total: 40000 W: 5175 L: 3430 D: 31395
40000 @ 60+0.6 th 1 Regression test until "simplify logic for history based pruning"
17-04-23 sg sort_tweak2 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 25925 W: 4688 L: 4756 D: 16481
sprt @ 10+0.1 th 1 weight 4250:break even at depth 17
17-04-23 sg excluded_no_update diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 24057 W: 4348 L: 4422 D: 15287
sprt @ 10+0.1 th 1 No stats update for excluded move
17-04-22 sg sort_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 77423 W: 13973 L: 13867 D: 49583
sprt @ 10+0.1 th 1 Less sorting: weight=3000. Test against passed sortLinearRebase patch.
17-04-22 sg sort_tweak diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 29589 W: 5458 L: 5201 D: 18930
sprt @ 10+0.1 th 1 Less sorting: shift by 4000. So we have for depth 1 the limit zero (the same as in current master). Test against passed sortLinearRebase patch.
17-04-21 sg opposite_castling diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18939 W: 3458 L: 3502 D: 11979
sprt @ 10+0.1 th 1 Less weight
17-04-20 sg opposite_castling diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 11615 W: 2141 L: 2216 D: 7258
sprt @ 10+0.1 th 1 Variant of Stephane's idea.
17-04-17 sg more_stats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 15466 W: 2789 L: 2848 D: 9829
sprt @ 10+0.1 th 1 Compensate history reduction 18000
17-04-17 sg more_stats diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10250 W: 1797 L: 1879 D: 6574
sprt @ 10+0.1 th 1 Compensate history reduction divisor 17000
17-04-17 sg more_stats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13822 W: 2503 L: 2569 D: 8750
sprt @ 10+0.1 th 1 Compensate history reduction. Take 2
17-04-17 sg more_stats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7763 W: 1347 L: 1439 D: 4977
sprt @ 10+0.1 th 1 Compensate history reduction
17-04-17 sg more_stats diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 17475 W: 3143 L: 3194 D: 11138
sprt @ 10+0.1 th 1 compensate update weights
17-04-17 sg more_stats diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16314 W: 2928 L: 2984 D: 10402
sprt @ 10+0.1 th 1 Extend stats bonus to depth 22
17-04-15 sg lmr diff
LLR: -2.93 (-2.94,2.94) [0.00,4.00]
Total: 30748 W: 5470 L: 5521 D: 19757
sprt @ 10+0.1 th 1 Strangly both tests failed yellow. So go further in both directions: Even more lmr correction
17-04-15 sg lmr2 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 28173 W: 5088 L: 5148 D: 17937
sprt @ 10+0.1 th 1 Even less lmr correction
17-04-14 sg lmr diff
LLR: -2.98 (-2.94,2.94) [0.00,4.00]
Total: 62864 W: 11357 L: 11301 D: 40206
sprt @ 10+0.1 th 1 more lmr correction
17-04-14 sg lmr2 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 54112 W: 9815 L: 9787 D: 34510
sprt @ 10+0.1 th 1 less lmr correction
17-04-14 sg update_cm_stats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18502 W: 3357 L: 3403 D: 11742
sprt @ 10+0.1 th 1 update counter move stats also for ss-3
17-04-13 sg smp_excluded_move diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6455 W: 986 L: 1083 D: 4386
sprt @ 5+0.05 th 7 Let some threads exclude the best move at root to look closer at the side variations, but only if best move is no capture or promotion.
17-04-12 sg smp_excluded_move diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9193 W: 1424 L: 1511 D: 6258
sprt @ 5+0.05 th 7 Let some threads exclude the best move at root to look closer at the side variations.
17-04-11 sg filtered_stats diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8067 W: 1405 L: 1496 D: 5166
sprt @ 10+0.1 th 1 Higher depth update have less noise than lower depth updates. So smoothing former less seems sensible. Alpha ranges now from 1/8 to 1.
17-04-10 sg filtered_stats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5269 W: 928 L: 1031 D: 3310
sprt @ 10+0.1 th 1 Even more smoothing
17-04-10 sg filtered_stats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10597 W: 1874 L: 1954 D: 6769
sprt @ 10+0.1 th 1 More smoothing
17-04-10 sg filtered_stats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8664 W: 1535 L: 1623 D: 5506
sprt @ 10+0.1 th 1 Less smoothing
17-04-10 sg filtered_stats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7263 W: 1233 L: 1327 D: 4703
sprt @ 10+0.1 th 1 Smooth history stats with an alpha/beta filter (a simplified kalman filter).
17-03-31 sg singular diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10895 W: 1923 L: 2002 D: 6970
sprt @ 10+0.1 th 1 Allow less singular extension on higher depth by using factor 9/4 for depth but compensate a little bit.Test against VoyagerOne's passed excludedTT patch.
17-03-31 sg singular diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18577 W: 3361 L: 3407 D: 11809
sprt @ 10+0.1 th 1 Allow more singular extension on higher depth by using factor 3/2 for depth but compensate a little bit.Test against VoyagerOne's passed excludedTT patch.
17-03-30 sg singular diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 33611 W: 6033 L: 6016 D: 21562
sprt @ 10+0.1 th 1 Allow less singular extension on higher depth by using factor 5/2 for depth but compensate a little bit.Test against VoyagerOne's passed excludedTT patch.
17-03-30 sg singular diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 9582 W: 1741 L: 1825 D: 6016
sprt @ 10+0.1 th 1 Allow more singular extension on higher depth by using factor 1 for depth but compensate a little bit.Test against VoyagerOne's passed excludedTT patch
17-03-30 sg singular diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 8482 W: 1469 L: 1559 D: 5454
sprt @ 10+0.1 th 1 Allow less singular extension on higher depth by using factor 3 for depth but compensate a little bit.Test against VoyagerOne's passed excludedTT patch.
17-03-27 sg singular diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 33527 W: 6056 L: 6039 D: 21432
sprt @ 10+0.1 th 1 Allow early pruning in singular search except futility pruning. Test against VoyagerOne's passed excludedTT patch.
17-03-27 sg singular diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9288 W: 1636 L: 1722 D: 5930
sprt @ 10+0.1 th 1 Allow early pruning in singular search but for compensation increase depth. Test against VoyagerOne's passed excludedTT patch.
17-03-27 sg singular diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10282 W: 1803 L: 1885 D: 6594
sprt @ 10+0.1 th 1 Allow early pruning in singular search except nullmove pruning. Test against VoyagerOne's passed excludedTT patch.
17-03-27 sg singular diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11746 W: 2112 L: 2187 D: 7447
sprt @ 10+0.1 th 1 Allow early pruning in singular search except iid. Test against VoyagerOne's passed excludedTT patch.
17-03-27 sg singular diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 39876 W: 7200 L: 7221 D: 25455
sprt @ 10+0.1 th 1 Allow early pruning in singular search. Test against VoyagerOne's passed excludedTT patch.
17-03-27 sg singular diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12265 W: 2134 L: 2207 D: 7924
sprt @ 10+0.1 th 1 Allow early pruning in singular search except probcut. Test against VoyagerOne's passed excludedTT patch.
17-03-26 sg update_stats4 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 157462 W: 28731 L: 28350 D: 100381
sprt @ 10+0.1 th 1 Test now 3/16 solution which is in the middle of best attempts
17-03-26 sg stats_bonus diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 48799 W: 8699 L: 8691 D: 31409
sprt @ 10+0.1 th 1 First test failed fast, so try opposite: stats bonus + 1
17-03-24 sg update_stats diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 40998 W: 7216 L: 7235 D: 26547
sprt @ 10+0.1 th 1 Half for all history stats the resolution.
17-03-25 sg update_stats3 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 174899 W: 31910 L: 31469 D: 111520
sprt @ 10+0.1 th 1 Last test struggles, so try an eighth resolution.
17-03-24 sg update_stats2 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 56416 W: 10138 L: 10103 D: 36175
sprt @ 10+0.1 th 1 My first two tests indicate that lower resolution is better, so go further and try quarter resolution. Put other test on prio -1.
17-03-26 sg stats_bonus diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 10021 W: 1755 L: 1877 D: 6389
sprt @ 10+0.1 th 1 My resolution series has the intention to lower the noise in computed stats (shows some very little gain so far). Now i try to go from the opposite site and filter the noise of the update values by decreasing the stats bonus by 1. For depth 1 (the noisiest one) we get then a bonus of zero, which avoids stats updates.
17-03-25 sg update_stats4 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 23034 W: 4166 L: 4244 D: 14624
sprt @ 10+0.1 th 1 Go now to the extreme 1/32 resolution (update weight 1).
17-03-25 sg update_stats4 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 25366 W: 4478 L: 4549 D: 16339
sprt @ 10+0.1 th 1 1/32 seems to low so try 1/16 resolution