Stockfish Testing Queue

Finished - 30472 tests

13-09-08 ren my_change diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 5935 W: 1169 L: 1260 D: 3506
sprt @ 15+0.05 th 1 Reduce scale factor for opposite colors bishop.
13-09-08 jki weak2 diff
ELO: -186.57 +-3.8 (95%) LOS: 0.0%
Total: 20000 W: 1095 L: 10909 D: 7996
20000 @ 15+0.05 th 1 weak vs. contempt 7
13-09-08 jki weak2 diff
ELO: -186.68 +-3.8 (95%) LOS: 0.0%
Total: 20000 W: 1136 L: 10955 D: 7909
20000 @ 15+0.05 th 1 weak vs. contempt 15
13-09-08 jki weak2 diff
ELO: -192.01 +-3.9 (95%) LOS: 0.0%
Total: 20000 W: 1120 L: 11170 D: 7710
20000 @ 15+0.05 th 1 weak vs. contempt 23
13-09-08 jki gpcont diff
ELO: -1.09 +-3.0 (95%) LOS: 24.1%
Total: 20000 W: 3981 L: 4044 D: 11975
20000 @ 15+0.05 th 1 Test game phase contempt
13-09-08 mco search diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 43999 W: 8807 L: 8792 D: 26400
sprt @ 15+0.05 th 1 Increase history ageing: take 1
13-09-08 mco search diff
LLR: -1.38 (-2.94,2.94) [-1.50,4.50]
Total: 42941 W: 8667 L: 8598 D: 25676
sprt @ 15+0.05 th 1 Increase history ageing: take 2
13-09-08 jos weak2 diff
ELO: -191.22 +-3.9 (95%) LOS: 0.0%
Total: 20000 W: 1236 L: 11252 D: 7512
20000 @ 15+0.05 th 1 weak2 vs. adaptive_contempt
13-09-08 lbr simplify_unstoppable diff
LLR: -2.97 (-2.94,2.94) [-3.00,3.00]
Total: 10233 W: 2035 L: 2143 D: 6055
sprt @ 15+0.05 th 1 simplify unstoppable passer eval
13-09-08 uri slightly_less_reduction diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 39414 W: 7887 L: 7885 D: 23642
sprt @ 15+0.05 th 1 additional try to solve 6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98 This time I only reduce second checks with evaluation that is bigger than 200. It means that in the line Rb3 axb3 Qe1+ Kg2 Qe4+ the move Qe4+ is not reduced because the evaluation before black played Qe4+ was clearly less than 2 pawns for black.
13-09-08 ren my_change diff
ELO: -15.30 +-13.4 (95%) LOS: 1.2%
Total: 1000 W: 171 L: 215 D: 614
1000 @ 15+0.05 th 1 Test scaling factor for rooks and queens endgame.
13-09-08 jki gcont diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 9024 W: 1846 L: 1929 D: 5249
sprt @ 15+0.05 th 1 Game phase contempt 7.5cp (bug fixed version)
13-09-08 pec combo2 diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 5962 W: 1140 L: 1230 D: 3592
sprt @ 15+0.05 th 1 Combo of 2 recent patches which failed at stage II with positive score.
13-09-08 jki noLowTT diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 45275 W: 9038 L: 9020 D: 27217
sprt @ 15+0.05 th 1 Do not access TT during recaptures
13-09-08 lbr noLowTT diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 4121 W: 935 L: 814 D: 2372
sprt @ 10+0.05 th 1 Do not access TT during recaptures. Reschedule in 10+0.05 with Hash=2. Consistent with 5'+3" Hash=64.
13-09-09 uri less_reduction2 diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 33876 W: 6907 L: 6706 D: 20263
sprt @ 15+0.05 th 1 another try to solve 6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98 faster that combine the best patch that failed with a positive score at stage 1 with more extensions for safe checks when the side that gives check has some disadvantage. stockfish with the new try solves the relevant position at depth 17.
13-09-09 lbr noLowTT diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 1394 W: 340 L: 229 D: 825
sprt @ 20+0.05 th 1 Do not access TT during recaptures. Passed SPRT(-1.5,4.5) 10+0.05 hash=2 Running SPRT(-1.5,4.5) 20+0.05 hash=4
13-09-09 uri less_reduction2 diff
LLR: -3.00 (-2.94,2.94) [-1.50,4.50]
Total: 20753 W: 4148 L: 4199 D: 12406
sprt @ 15+0.05 th 1 another try to solve 6k1/1r4p1/2p2Pb1/2p3R1/p1Ppq3/Q7/PP3R2/6K1 b - - 2 98 with an improvement. This time I added some risky reduction when the evaluation is high for one of the players so it can fail fast or pass fast.
13-09-09 jki gcont diff
ELO: 0.66 +-3.1 (95%) LOS: 66.4%
Total: 20000 W: 4064 L: 4026 D: 11910
20000 @ 15+0.05 th 1 Game phase contempt 7.5cp (bug fixed version), fixed games
13-09-09 mco search diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 13093 W: 2723 L: 2579 D: 7791
sprt @ 15+0.05 th 1 Increase history ageing: take 3
13-09-09 mco search diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 5407 W: 1134 L: 1012 D: 3261
sprt @ 15+0.05 th 1 Increase history ageing: take 4
13-09-09 lbr noLowTT diff
LLR: -1.65 (-2.94,2.94) [-1.50,4.50]
Total: 58800 W: 12303 L: 12196 D: 34301
sprt @ 10+0.05 th 1 Do not access TT during recaptures. Mistake while creating test. Idea is to test again with a realistic hash p[ressure: 10+0.05 hash=2, consistent with 5'+3" hash=64 (or hash=128 with pondering).
13-09-09 uri less_reduction2 diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 15278 W: 3116 L: 2967 D: 9195
sprt @ 15+0.05 th 1 similiar to the patch that passed stage 1 but this time I extend by one ply more checks and all checks when the evaluation is smaller than zero before the checking move get extended by one ply.
13-09-09 lbr qs_fut_pv diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 57000 W: 11253 L: 11204 D: 34543
sprt @ 15+0.05 th 1 QS Futility pruning at PV nodes
13-09-09 uri less_reduction2 diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 20262 W: 4020 L: 4071 D: 12171
sprt @ 15+0.05 th 1 another try to extend more checks in special cases before I decide what to commit for long time control.
13-09-09 uri search diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 81481 W: 16363 L: 16245 D: 48873
sprt @ 15+0.05 th 1 idea of marco and I push another test
13-09-09 lbr history diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 2910 W: 549 L: 649 D: 1712
sprt @ 15+0.05 th 1 half history table if overflow
13-09-09 rap threat diff
LLR: -2.99 (-2.94,2.94) [-1.50,4.50]
Total: 4761 W: 906 L: 1001 D: 2854
sprt @ 15+0.05 th 1 Bonus attacks of pawn defended pieces.
13-09-09 jki gcont23.5 diff
ELO: 200.79 +-4.0 (95%) LOS: 100.0%
Total: 20000 W: 11569 L: 1146 D: 7285
20000 @ 15+0.05 th 1 game phase contempt (23.5) vs. weak2
13-09-09 uri less_reduction2 diff
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 24228 W: 4233 L: 4012 D: 15983
sprt @ 60+0.05 th 1 The same test that passed short time control best in long time control and note that 2 tests passed short time control.
13-09-09 pec search diff
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 19870 W: 3339 L: 3346 D: 13185
sprt @ 60+0.05 th 1 LTC of best attempt for Marco. Increase history ageing: take 4
13-09-09 mco search diff
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 21021 W: 3622 L: 3622 D: 13777
sprt @ 60+0.05 th 1 LTC: Increase history ageing: take 3
13-09-10 lbr reduction diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 25106 W: 5073 L: 5110 D: 14923
sprt @ 15+0.05 th 1 reduce lextended moves one ply less
13-09-10 Fis fast_tt_access diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 20436 W: 4283 L: 4117 D: 12036
sprt @ 10+0.05 th 1 Optimized TranspositionTable::store() to: 1) do less work when finding replacement entries. 2) store most recent entries such that TranspositionTable::probe() finds them faster NOTE: same as the "noLowTT" test/branch I'm setting Hash low to simulate realistic hash pressure.
13-09-10 jki gcont23.5 diff
ELO: -2.47 +-3.3 (95%) LOS: 7.1%
Total: 20000 W: 4603 L: 4745 D: 10652
20000 @ 15+0.05 th 1 game phase contempt (23.5) vs. master
13-09-10 mco uri_ext diff
LLR: 2.98 (-2.94,2.94) [-1.50,4.50]
Total: 3974 W: 860 L: 741 D: 2373
sprt @ 15+0.05 th 1 Bisect Uri extension patch: take 1
13-09-10 rap passed_A diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 30126 W: 5963 L: 5987 D: 18176
sprt @ 15+0.05 th 1 Add bonus for passed pawn blocked by friendly piece
13-09-10 mco unstoppable diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 6636 W: 1288 L: 1377 D: 3971
sprt @ 15+0.05 th 1 Another attempt at removing unstoppable evaluation. This time the idea is to use candiates and not passed as potential unstoppable.
13-09-10 lbr reduction diff
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 42466 W: 8609 L: 8385 D: 25472
sprt @ 15+0.05 th 1 reduce checks one ply less
13-09-10 mco search diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 16326 W: 2784 L: 2807 D: 10735
sprt @ 60+0.05 th 1 LTC: Increase history ageing: take 1 (this failed at short TC but other tests proved this tweak to be very TC sensitive and showing less increase is better at long TC).
13-09-10 rap threat_A diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 17732 W: 3493 L: 3551 D: 10688
sprt @ 15+0.05 th 1 Give bonus for piece threat on same type of piece.
13-09-10 uri uri_ext diff
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 16807 W: 2917 L: 2733 D: 11157
sprt @ 60+0.05 th 1 Bisect Uri extension patch: take 1
13-09-10 mco unstoppable diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 46499 W: 9356 L: 9334 D: 27809
sprt @ 15+0.05 th 1 Second attempt at unstoppable
13-09-10 rap king_margin diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 9177 W: 1841 L: 1923 D: 5413
sprt @ 15+0.05 th 1 Reduce king safety margins.
13-09-10 Fis fast_tt_access diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 52019 W: 10505 L: 10468 D: 31046
sprt @ 15+0.05 th 1 Optimized TranspositionTable::store() to: 1) do less work when finding replacement entries. 2) store most recent entries such that TranspositionTable::probe() finds them faster. Run #2 at usual TC and hash sizes per Marcos suggestion.
13-09-11 lbr check_ext diff
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 42630 W: 8684 L: 8342 D: 25604
sprt @ 15+0.05 th 1 try to simplify Uri's extension: extend checks by ONE_PLY.
13-09-11 mco uri_ext diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 8739 W: 1830 L: 1698 D: 5211
sprt @ 15+0.05 th 1 Check extension: Compare with alpha instead of VALUE_ZERO
13-09-11 lbr check_ext diff
LLR: -0.07 (-2.94,2.94) [0.00,6.00]
Total: 445 W: 82 L: 82 D: 281
sprt @ 60+0.05 th 1 try to simplify Uri's extension: extend checks by ONE_PLY.
13-09-11 lbr uri_ext diff
LLR: 2.96 (-2.94,2.94) [0.00,6.00]
Total: 6716 W: 1238 L: 1101 D: 4377
sprt @ 60+0.05 th 1 Check extension: Compare with alpha instead of VALUE_ZERO
13-09-11 rap king_tune_A diff
ELO: -0.07 +-3.0 (95%) LOS: 48.2%
Total: 20000 W: 3990 L: 3994 D: 12016
20000 @ 15+0.05 th 1 Tune king safety