Stockfish Testing Queue

Finished - 24195 tests

28-05-14 aj constricted_bishop3 diff
LLR: -1.13 (-2.94,2.94) [-1.50,4.50]
Total: 2708 W: 479 L: 512 D: 1717
sprt @ 15+0.05 th 1 Reduce pawn penalties if bishop is not restricted to only one side of the board: STC : final take
27-05-14 ra see_prune diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 43263 W: 7592 L: 7588 D: 28083
sprt @ 15+0.05 th 1 Prune very negative SEE moves at higher depth
27-05-14 lb symks diff
ELO: -1.97 +-2.0 (95%) LOS: 2.6%
Total: 40000 W: 6681 L: 6908 D: 26411
40000 @ 15+0.05 th 1 symmetric ks with spsa tuned weight. measure at different tc.
27-05-14 lb symks diff
ELO: -3.03 +-2.2 (95%) LOS: 0.3%
Total: 40000 W: 7870 L: 8219 D: 23911
40000 @ 6+0.02 th 1 symmetric ks with spsa tuned weight. measure at different tc.
27-05-14 aj constricted_bishop3 diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 3849 W: 612 L: 706 D: 2531
sprt @ 15+0.05 th 1 Reduce pawn penalties if bishop is not restricted to only one side of the board: STC : Take 3
26-05-14 mc 5e03734eacc8e52a6c92b diff
ELO: 55.26 +-2.2 (95%) LOS: 100.0%
Total: 40000 W: 11417 L: 5108 D: 23475
40000 @ 60+0.05 th 1 Regression test just before symmetric king safety
26-05-14 lb master diff
ELO: 53.28 +-2.1 (95%) LOS: 100.0%
Total: 40000 W: 10944 L: 4858 D: 24198
40000 @ 60+0.05 th 1 Regression test until 495a0fa6 (2014-05-25). Low priority, only because framework empty.
27-05-14 Fi tt_search diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 21863 W: 3839 L: 3888 D: 14136
sprt @ 15+0.05 th 1 Optimize TT search inside probe() and store(). See commit for explanation.
27-05-14 lb 5e03734e diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 74394 W: 13047 L: 13015 D: 48332
sprt @ 15+0.05 th 1 90% probability of regression here! These 2 patches have NOT been tested correctly (stopped SPRT runs, STC without LTC...)
26-05-14 jh singular_ext diff
LLR: -2.97 (-2.94,2.94) [0.00,6.00]
Total: 4113 W: 534 L: 617 D: 2962
sprt @ 60+0.05 th 1 Only extend when ttValue > alpha. LTC for more extension hits.
26-05-14 aj constricted_bishop diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 9268 W: 1483 L: 1564 D: 6221
sprt @ 15+0.05 th 1 Reduce pawn penalties if bishop is not restricted to only one side of the board: STC : Take 2
26-05-14 ra threat_B diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 14922 W: 2417 L: 2484 D: 10021
sprt @ 15+0.05 th 1 More LMR for not avoiding threat.
26-05-14 ra threat diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 10849 W: 1772 L: 1849 D: 7228
sprt @ 15+0.05 th 1 Less LMR for moves that avoiding threats
26-05-14 aj constricted_bishop diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 32213 W: 5511 L: 5535 D: 21167
sprt @ 15+0.05 th 1 Reduce pawn penalties if bishop is not restricted to only one side of the board: STC : Take 1
26-05-14 jo null1 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 2640 W: 396 L: 493 D: 1751
sprt @ 15+0.05 th 1 No null-move at shallow depths.
23-05-14 Fi master diff
ELO: -2.19 +-2.0 (95%) LOS: 1.4%
Total: 40000 W: 6558 L: 6810 D: 26632
40000 @ 15+0.05 th 1 Test to determine any elo difference between using 8MB vs 128MB hash at STC. Low priority. I hope this finally hits an elo drop!
25-05-14 Fi master diff
ELO: -13.91 +-2.0 (95%) LOS: 0.0%
Total: 40000 W: 5833 L: 7434 D: 26733
40000 @ 15+0.05 th 1 Test to determine any elo difference between using 4MB vs 128MB hash at STC. Low priority. Last number I need before I can post a forum summary.
25-05-14 ca patch-1 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 9213 W: 1528 L: 1609 D: 6076
sprt @ 15+0.05 th 1
25-05-14 ra mate_threat diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 15993 W: 2665 L: 2729 D: 10599
sprt @ 15+0.05 th 1 Use mate threat detected by null move
25-05-14 jh null_verification diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 48968 W: 8070 L: 8303 D: 32595
sprt @ 15+0.05 th 1 Always do null verification.
25-05-14 lb tune_lmr diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 19601 W: 2559 L: 2734 D: 14308
sprt @ 60+0.05 th 1 LTC: simplified LMR tuned by SPSA. Failed SPRT(-3,1) but survived for a while, and scaling could come into play here.
23-05-14 in mega_pawn_tune diff
72665/100000 iterations
200000/200000 games played
200000 @ 15+0.05 th 1 Tuning doubled, isolated, backward, pawn_file_span, unsupported. (ck=8) // This should keep the framework busy for some while... (44 params total).
25-05-14 My weakEn diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 41957 W: 6973 L: 6975 D: 28009
sprt @ 15+0.05 th 1 Pieces attacked by enemy king are also weakenemies.
25-05-14 mc dense_tt diff
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 10917 W: 1461 L: 1516 D: 7940
sprt @ 60+0.05 th 1 LTC: Revert to master and set ClusterSize = 3
25-05-14 mc dense_tt^ diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 8827 W: 1160 L: 1223 D: 6444
sprt @ 60+0.05 th 1 LTC: Pad to cache line and set ClusterSize = 3
25-05-14 mc dense_tt diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 38030 W: 6425 L: 6231 D: 25374
sprt @ 15+0.05 th 1 Revert to master and set ClusterSize = 3
25-05-14 mc dense_tt^ diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 18283 W: 3139 L: 2992 D: 12152
sprt @ 15+0.05 th 1 Pad to cache line and set ClusterSize = 3
25-05-14 mc dense_tt diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 4007 W: 608 L: 701 D: 2698
sprt @ 15+0.05 th 1 Ensure cluster is cache line aligned
23-05-14 Fi master diff
ELO: 0.44 +-2.0 (95%) LOS: 67.1%
Total: 40000 W: 6662 L: 6611 D: 26727
40000 @ 15+0.05 th 1 Test to determine any elo difference between using 16MB vs 128MB hash at STC. Low priority.
24-05-14 lb tune_lmr diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 77357 W: 12649 L: 12926 D: 51782
sprt @ 15+0.05 th 1 LMR 3 param: tuned by SPSA.
24-05-14 mc dense_tt^ diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 63723 W: 10720 L: 10669 D: 42334
sprt @ 15+0.05 th 1 Set ClusterSize = 3
24-05-14 mc dense_tt diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 6619 W: 1022 L: 1109 D: 4488
sprt @ 15+0.05 th 1 Set ClusterSize = 5
24-05-14 jh cheap_null_verification diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 21754 W: 2974 L: 2984 D: 15796
sprt @ 60+0.05 th 1 LTC: Always do verification when depth-R <= ONE_PLY.
23-05-14 lb tune_lmr diff
75258/100000 iterations
159842/200000 games played
200000 @ 15+0.05 th 1 3 param LMR (reintroduce improving)
24-05-14 mc dense_tt diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 5910 W: 964 L: 1053 D: 3893
sprt @ 15+0.05 th 1 Compress TTEntry by 15% without lose of precision (but a slowdown).
24-05-14 jh cheap_null_verification diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 13381 W: 2305 L: 2170 D: 8906
sprt @ 15+0.05 th 1 Always do verification at depth < ONE_PLY.
24-05-14 mc inline__tt_save diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 11049 W: 1769 L: 1819 D: 7461
sprt @ 15+0.05 th 1 Test a possible speed optimization (seems faster locally)
24-05-14 in mega_pawn diff
ELO: 0.90 +-2.3 (95%) LOS: 78.0%
Total: 30000 W: 5145 L: 5067 D: 19788
30000 @ 15+0.05 th 1 Check progress of mega_pawn_tune after 48K iterations.
22-05-14 Fi master diff
ELO: 2.59 +-2.0 (95%) LOS: 99.5%
Total: 40000 W: 6800 L: 6502 D: 26698
40000 @ 15+0.05 th 1 Test to determine any elo difference between using 128MB vs 32MB hash at STC. Low priority.
23-05-14 lb tune_lmr diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 71026 W: 11627 L: 11894 D: 47505
sprt @ 15+0.05 th 1 simplified LMR tuned by SPSA
22-05-14 lb tune_lmr diff
74591/75000 iterations
157707/175000 games played
175000 @ 15+0.05 th 1 Tune LMR formula in simplified form (use ck=8).
23-05-14 sg lmr2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 14767 W: 2443 L: 2510 D: 9814
sprt @ 15+0.05 th 1 more LMR if in check and history < 0. Condition selected based on collected stats (11% less fail highs).
23-05-14 sn little_combinaisons7 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 17813 W: 2917 L: 2977 D: 11919
sprt @ 15+0.05 th 1 Anticipate winning exchange sequences on weak pawns.
23-05-14 mc tt_dense diff
LLR: -0.02 (-2.94,2.94) [0.00,6.00]
Total: 233 W: 26 L: 26 D: 181
sprt @ 60+0.05 th 3 Real LTC: 3 threads to simulate LTC but use less RAM
22-05-14 Ro tt_dense diff
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 13289 W: 2130 L: 1969 D: 9190
sprt @ 20+0.05 th 3 3 threads to simulate LTC but use less RAM
22-05-14 gl StandPat_a diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 4058 W: 619 L: 712 D: 2727
sprt @ 15+0.05 th 1 Reschedule for EE: Avoid returning inferior depth beta cutoffs
22-05-14 mc spsa_mega_tuning diff
LLR: -0.68 (-2.94,2.94) [0.00,4.00]
Total: 45950 W: 6398 L: 6306 D: 33246
sprt @ 60+0.05 th 1 Further increase king safety
22-05-14 in stormdanger diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 21738 W: 3032 L: 3121 D: 15585
sprt @ 60+0.05 th 1 LTC [4..0]: Tuned values from SPSA.
22-05-14 Ro tt_dense diff
LLR: -0.32 (-2.94,2.94) [0.00,6.00]
Total: 677 W: 146 L: 154 D: 377
sprt @ 60+0.05 th 1 LTC: 512MB hash
22-05-14 in stormdanger diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 19654 W: 3454 L: 3302 D: 12898
sprt @ 15+0.05 th 1 Tuned values from SPSA.