Stockfish Testing Queue

Finished - 24195 tests

21-05-14 in tuning_stormdanger diff
47989/50000 iterations
100000/100000 games played
100000 @ 30+0.1 th 1 Tuning locally failed because I used very short TC. Try again with much longer TC as storm code is TC sensitive. (Corrected values)
22-05-14 ca master diff
8276/10000 iterations
16830/20000 games played
20000 @ 60+0.05 th 1
22-05-14 sn rook_passers diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 8048 W: 1375 L: 1459 D: 5214
sprt @ 15+0.05 th 1 Rook passers are strong when the defensive side does not have a double rook.
21-05-14 mc spsa_mega_tuning diff
LLR: -0.38 (-2.94,2.94) [0.00,4.00]
Total: 52391 W: 7335 L: 7210 D: 37846
sprt @ 60+0.05 th 1 Change only king safety, test at LTC with new SPRT(0, 4) setup
18-05-14 My con diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 51719 W: 7761 L: 7635 D: 36323
sprt @ 60+0.05 th 1 LTC: fixed contempt that only affects the side to move.
21-05-14 Ro tt_dense diff
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 25967 W: 3820 L: 3808 D: 18339
sprt @ 60+0.05 th 1 LTC: 1.5x denser TT
21-05-14 Ro tt_dense diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 7553 W: 1363 L: 1241 D: 4949
sprt @ 15+0.05 th 1 Pack values without using bitfields
20-05-14 lb spsa diff
LLR: -2.95 (-2.94,2.94) [-1.00,4.00]
Total: 79698 W: 13310 L: 13242 D: 53146
sprt @ 15+0.05 th 1 SPSA tuned values. Test with 5 bayeselo resolution: SPRT(-1,4) at STC and SPRT(0,5) at LTC, if STC passes.
21-05-14 in tuning_stormdanger diff
9610/50000 iterations
19754/100000 games played
100000 @ 30+0.1 th 1 Tuning locally failed because I used very short TC. Try again with much longer TC as storm code is TC sensitive.
21-05-14 in backward diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 13614 W: 2228 L: 2298 D: 9088
sprt @ 15+0.05 th 1 Condition for slightly rare cases when friendly pawn exists behind on adjacent files, but it cannot advance because its path is blocked.
21-05-14 Ro tt_dense diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 15783 W: 2761 L: 2825 D: 10197
sprt @ 15+0.05 th 1 22-bit entry key with 6 entries/cluster
20-05-14 gl spsa_mega_tuning^ diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 27050 W: 3791 L: 3865 D: 19394
sprt @ 60+0.05 th 1 Speculative LTC, SPRT[0..4], Test samples at plateau 1
20-05-14 mc spsa_mega_tuning^ diff
LLR: -4.34 (-2.94,2.94) [-1.50,4.50]
Total: 73307 W: 12340 L: 12314 D: 48653
sprt @ 15+0.05 th 1 Test samples at plateau 1
19-05-14 gl master diff
75844/50000 iterations
160000/160000 games played
160000 @ 30+0.1 th 1 Reschedule at 30+0.1: tune eval macro weights (5 param). only most sensitive ones. an interesting challenge for SPSA!
20-05-14 mc spsa_mega_tuning diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 17569 W: 2919 L: 2980 D: 11670
sprt @ 15+0.05 th 1 Test samples at plateau 2
19-05-14 lb master diff
69552/100000 iterations
168265/200000 games played
200000 @ 9+0.03 th 1 tune eval macro weights (5 param). only most sensitive ones. an interesting challenge for SPSA!
19-05-14 mc spsa_mega_tuning diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 8903 W: 1475 L: 1557 D: 5871
sprt @ 15+0.05 th 1 Verify SPSA tuning after 100K games (without king safety)
19-05-14 mc spsa_mega_tuning^ diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 33364 W: 5605 L: 5627 D: 22132
sprt @ 15+0.05 th 1 Verify SPSA tuning after 100K games
19-05-14 ra prob_B diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 4460 W: 706 L: 798 D: 2956
sprt @ 15+0.05 th 1 Reduced depth search for better captures. Take 2.
17-05-14 mc master diff
LLR: 1.27 (-2.94,2.94) [-3.00,1.00]
Total: 130152 W: 19023 L: 19146 D: 91983
sprt @ 60+0.05 th 1 LTC: Test for no-regression "Extract a reliable PV line" from an idea of Ronald de Man
19-05-14 ra prob_A diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 20874 W: 3655 L: 3707 D: 13512
sprt @ 15+0.05 th 1 Reduced depth search for better captures
17-05-14 sh onepiece diff
LLR: 4.27 (-2.94,2.94) [0.00,6.00]
Total: 95949 W: 14251 L: 13694 D: 68004
sprt @ 60+0.05 th 1 LTC: Scale one piece endgames based on pawn span.
19-05-14 Ro tt_dense diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 3919 W: 619 L: 713 D: 2587
sprt @ 15+0.05 th 1 1.5x denser TT with smaller hash for STC
17-05-14 gl simplify_passed diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 49848 W: 7468 L: 7693 D: 34687
sprt @ 60+0.05 th 1 LTC for Marco (framework empty again!): Retry a successful simplification from Arjun
19-05-14 in hole diff
LLR: -0.85 (-2.94,2.94) [-1.50,4.50]
Total: 536 W: 90 L: 119 D: 327
sprt @ 15+0.05 th 1 Smaller bonus for outpost, much higher for hole.
19-05-14 in hole diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 5592 W: 915 L: 1005 D: 3672
sprt @ 15+0.05 th 1 Even smaller bonus.
19-05-14 in hole diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 3611 W: 580 L: 675 D: 2356
sprt @ 15+0.05 th 1 Take 2. Slightly reduce bonus for outposts to compensate for additional bonus for "hole".
19-05-14 in hole diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 3078 W: 514 L: 611 D: 1953
sprt @ 15+0.05 th 1 Concept of hole.
18-05-14 sh onepiece diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 18536 W: 3179 L: 3237 D: 12120
sprt @ 15+0.05 th 1 Scale one piece endgames based on pawn span (take 2)
18-05-14 Ro tt_dense diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 14426 W: 2517 L: 2585 D: 9324
sprt @ 15+0.05 th 1 1.5x denser TT
18-05-14 sn pinned_pawns5 diff
LLR: -4.91 (-2.94,2.94) [-1.50,4.50]
Total: 26109 W: 4533 L: 4640 D: 16936
sprt @ 15+0.05 th 1 Experimental run: don't get a malus for a piece attacked by a pinned pawn
18-05-14 sn pinned_pawns5 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 8231 W: 1419 L: 1502 D: 5310
sprt @ 15+0.05 th 1 Experimental run: give a bonus for a piece attacked by a pinned pawn
18-05-14 My con diff
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 54912 W: 9851 L: 9609 D: 35452
sprt @ 15+0.05 th 1 With dynamic contempt over, attempt fixed contempt that only affects the side to move.
17-05-14 gl master diff
7420/50000 iterations
73592/150000 games played
150000 @ 15+0.05 th 1 SPSA trial run
18-05-14 lb symks diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 40571 W: 5852 L: 5760 D: 28959
sprt @ 60+0.05 th 1 symmetric king safety
18-05-14 ra unnecessary diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 1261 W: 190 L: 292 D: 779
sprt @ 15+0.05 th 1 Prune moves that threats moves that could be done before, like e2e3 threats e3e4.
18-05-14 lb symks diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 33855 W: 5863 L: 5764 D: 22228
sprt @ 15+0.05 th 1 symmetric king safety
17-05-14 mc master diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 122092 W: 21290 L: 21337 D: 79465
sprt @ 15+0.05 th 1 Quick test to verify last bug fix is ok
18-05-14 My razor_null diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 16583 W: 2868 L: 2931 D: 10784
sprt @ 15+0.05 th 1 Include null in razoring. Tested well locally.
18-05-14 in agg_cow diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 30288 W: 5330 L: 5358 D: 19600
sprt @ 15+0.05 th 1 Quick test of results obtained from SPRT trial run.
17-05-14 tk Queen_Rook_Outposts diff
LLR: -4.39 (-2.94,2.94) [-1.50,4.50]
Total: 12906 W: 2211 L: 2333 D: 8362
sprt @ 15+0.05 th 1 Different approach to Queen and Rook Outposts/Tropism from sn's patch.
16-05-14 pe contempt diff
ELO: 56.77 +-2.2 (95%) LOS: 100.0%
Total: 40000 W: 11491 L: 5013 D: 23496
40000 @ 60+0.05 th 1 Regression test after "Drop to qsearch at low depth in razoring" + dynamic contempt to measure effect separately
17-05-14 gl master diff
ELO: 56.55 +-1.9 (95%) LOS: 100.0%
Total: 60000 W: 18869 L: 9189 D: 31942
60000 @ 15+0.05 th 1 Low pri short TC regression test after razoring change, just to compare short TC vs long TC performance
17-05-14 jo npm_bonus diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 32565 W: 5745 L: 5767 D: 21053
sprt @ 15+0.05 th 1 Don't apply bonus if one side has the pair of bishops. Final take.
17-05-14 Fi tt_overwrite diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 6821 W: 1155 L: 1242 D: 4424
sprt @ 15+0.05 th 1 Don't overwrite matching TT entries w/ lower depth data. If this passes please LTC for me.
17-05-14 lb pawns diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4595 W: 752 L: 857 D: 2986
sprt @ 15+0.05 th 1 take 3
17-05-14 mc master diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 32180 W: 5684 L: 5582 D: 20914
sprt @ 15+0.05 th 1 Test for no-regression "Extract a reliable PV line" from an idea of Ronald de Man
15-05-14 gl master diff
ELO: 57.15 +-1.8 (95%) LOS: 100.0%
Total: 60000 W: 17070 L: 7289 D: 35641
60000 @ 60+0.05 th 1 Regression test after razoring change
17-05-14 lb pawns diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 27336 W: 4773 L: 4784 D: 17779
sprt @ 15+0.05 th 1 only rank2
17-05-14 lb pawns diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 3493 W: 588 L: 753 D: 2152
sprt @ 15+0.05 th 1 remove pawn attack on edge. seems a bit redundant with huge pawn storm bonus on 6th rank now.