Stockfish Testing Queue

Finished - 30521 tests

15-11-11 Voy hpk diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 13461 W: 2713 L: 2525 D: 8223
sprt @ 10+0.1 th 1 Don't history prune first killer move (fixed bench)
15-11-11 Voy USt3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 20844 W: 4052 L: 4085 D: 12707
sprt @ 10+0.1 th 1 Take 3- Don't penalize regular history at depth=1
15-11-11 bin passed_imbalance diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13716 W: 2639 L: 2705 D: 8372
sprt @ 10+0.1 th 1 King proximity to passed pawns is vital in imbalances
15-11-11 wf movepick-score-tune diff
4689/5000 iterations
9397/10000 games played
10000 @ 5+0.05 th 3 SMP tuning experiment: fast TC and th=3 for many games, then subsequent small game, small value range tuning pass with th=7. Please see branch commit for comments. (Take 2: ck=20 is delicious.)
15-11-11 sni tune_threat2 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 29966 W: 5781 L: 5832 D: 18353
sprt @ 10+0.1 th 1 Try tuned threat values (resubmit with correct [0..4] bounds)
15-11-10 luc MoveEasiness diff
ELO: -2.81 +-3.0 (95%) LOS: 3.1%
Total: 20000 W: 3719 L: 3881 D: 12400
20000 @ 10+0.1 th 1 Easy move handled in normal TM, also stackable: preliminary Elo impact measurement @ STC
15-11-11 wf movepick-score-tune diff
692/5000 iterations
1371/10000 games played
10000 @ 5+0.05 th 3 SMP tuning experiment: fast TC and th=3 for many games, then subsequent small game, small value range tuning pass with th=7. Please see branch commit for comments.
15-11-10 sni tune_threat diff
38386/40000 iterations
79900/80000 games played
80000 @ 20+0.2 th 1 Tune threat array
15-11-11 sni tune_threat2 diff
LLR: 0.29 (-2.94,2.94) [0.00,5.00]
Total: 2672 W: 509 L: 485 D: 1678
sprt @ 10+0.1 th 1 Try tuned threat values
15-11-11 Voy USt2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18341 W: 3504 L: 3549 D: 11288
sprt @ 10+0.1 th 1 Take 2. Don't penalize moves if depth <3.
15-11-11 Roc XDouble diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17841 W: 3399 L: 3446 D: 10996
sprt @ 10+0.1 th 1 Take 2: consider only captures by Minors
15-11-10 tvi LMR diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 21565 W: 3733 L: 3768 D: 14064
sprt @ 5+0.05 th 7 Try idx / 2
15-11-11 Roc XDouble diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9885 W: 1902 L: 1985 D: 5998
sprt @ 10+0.1 th 1 Threat to create a double pawn
15-11-10 Voy USt diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 59104 W: 11389 L: 11253 D: 36462
sprt @ 10+0.1 th 1 Update State tweak- Don't penalize moves when depth=1
15-11-10 Voy StatPenalty-2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16211 W: 3084 L: 3139 D: 9988
sprt @ 10+0.1 th 1 Take 2
15-11-10 tvi LMR diff
LLR: -2.25 (-2.94,2.94) [0.00,5.00]
Total: 20000 W: 3533 L: 3544 D: 12923
sprt @ 5+0.05 th 7 More LMR for slaves
15-11-10 Voy StatPenalty diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5147 W: 905 L: 1009 D: 3233
sprt @ 10+0.1 th 1 Smaller stat penalty
15-11-10 jos development diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13193 W: 2463 L: 2531 D: 8199
sprt @ 10+0.1 th 1 Take 2. Explicitly exclude king and rooks. Also slightly increased bonus.
15-11-10 jos development diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 24989 W: 4777 L: 4793 D: 15419
sprt @ 10+0.1 th 1 Give a small penalty for each piece on the back-rank (except king and 2 rooks) during opening.
15-11-10 roh helper-thread-depth-tun diff
4625/10000 iterations
9230/20000 games played
20000 @ 20+0.2 th 7 Tuning run for the test that has passed at STC & LTC. Rescheduled after removing nodestime.
15-11-10 sg outposts2 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 30507 W: 4428 L: 4491 D: 21588
sprt @ 60+0.4 th 1 LTC: The changes are small but test tuned outpost weights. Perhaps a further tuning session is needed. Test against passed possible outpost patch.
15-11-10 Voy StatIdea diff
LLR: -2.61 (-2.94,2.94) [0.00,5.00]
Total: 3428 W: 621 L: 718 D: 2089
sprt @ 10+0.1 th 1 Another Stat Idea...
15-11-10 jos pawn_queen_psqt diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 6968 W: 1364 L: 1499 D: 4105
sprt @ 10+0.1 th 1 Another tuning session on pawn and queen psqt.
15-11-10 sg king_safety diff
LLR: -3.04 (-2.94,2.94) [0.00,5.00]
Total: 7950 W: 1453 L: 1548 D: 4949
sprt @ 10+0.1 th 1 Combine two of my yellow king safety patches: 1. unsafe queen contact checks 2. pinned king attacks (this only rearrange code) http://tests.stockfishchess.org/tests/view/563ca0d00ebc59227df1700f http://tests.stockfishchess.org/tests/view/5637d8990ebc59227df16e14 Lower throughput
15-11-10 sg outposts2 diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 20455 W: 4058 L: 3826 D: 12571
sprt @ 10+0.1 th 1 The changes are small but test tuned outpost weights. Perhaps a further tuning session is needed. Test against passed possible outpost patch.
15-11-10 tvi LMR diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 2550 W: 425 L: 540 D: 1585
sprt @ 5+0.05 th 7 Less LMR for slaves, new STC 5+0.05
15-11-09 sg tune_outposts diff
37652/40000 iterations
80000/80000 games played
80000 @ 20+0.2 th 1 Tune outpost stuff after my passed possible outpost patch.
15-11-10 Roc RookOnOpen diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7142 W: 1334 L: 1429 D: 4379
sprt @ 10+0.1 th 1 Bonus for Rook which can reach a better file. (Fixed bug)
15-11-10 Roc RookOnOpen diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 4569 W: 835 L: 942 D: 2792
sprt @ 10+0.1 th 1 Allocate bonus only if potential improving square is free and not attacked by Minors
15-11-09 pec not_failedlow diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12488 W: 1834 L: 1909 D: 8745
sprt @ 60+0.4 th 1 Increment bestMoveChanges only by half if not failed low at root. For STC it was likely small positive change, so I want to see how it does at LTC -- using priority -1, so it will run when there is no one threaded jobs
15-11-10 Voy statsIdea diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 26692 W: 5178 L: 5185 D: 16329
sprt @ 10+0.1 th 1 Try out a stat idea...
15-11-09 roh helper-thread-depth diff
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 21467 W: 2693 L: 2510 D: 16264
sprt @ 60+0.4 th 7 Reduce helper thread depth when running out of available time. Passed at STC. Now testing at LTC.
15-11-09 Ser prev_moves_3fold diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 34676 W: 5205 L: 5199 D: 24272
sprt @ 60+0.4 th 1 LTC. Passed STC in http://tests.stockfishchess.org/tests/view/563fb46f0ebc59227df1710b Code comments changed, otherwise same code.
15-11-09 SC variance3 diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 1003 W: 131 L: 252 D: 620
sprt @ 10+0.1 th 1 I think I have badly screwed my tuning session. But let us test anyway whether I have gone somewhere meaningful.
15-11-09 wf movepick-score-st-depth diff
LLR: -1.68 (-2.94,2.94) [0.00,5.00]
Total: 7913 W: 1246 L: 1285 D: 5382
sprt @ 10+0.1 th 7 Choose a high scoring, nearly-deepest thread.
15-11-09 Voy StatUpdateT diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 2797 W: 481 L: 596 D: 1720
sprt @ 10+0.1 th 1 Stat update tweak (Fix bench)
15-11-09 Fis easy2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18750 W: 3600 L: 3643 D: 11507
sprt @ 10+0.1 th 1 Allow a 2nd easy move if it's an easy recapture. (There probably isn't enough of these to make a difference and pass but let's see.) Bug fixed version.
15-11-08 jki timeman_log diff
ELO: 2.26 +-2.8 (95%) LOS: 94.1%
Total: 20000 W: 3524 L: 3394 D: 13082
20000 @ 1+1 th 1 Timeman log-tweak, high increment games, check ELO impact
15-11-09 lbr test diff
LLR: 0.21 (-2.94,2.94) [-3.00,1.00]
Total: 7419 W: 1420 L: 1422 D: 4577
sprt @ 10+0.1 th 1 fix bestThread
15-11-09 Fis easy2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 20274 W: 3829 L: 3866 D: 12579
sprt @ 10+0.1 th 1 Allow a 2nd easy move if it's an easy recapture. (There probably isn't enough of these to make a difference and pass but let's see.)
15-11-08 SC varAspTuning diff
19607/20000 iterations
40000/40000 games played
40000 @ 10+0.1 th 1 The two last variance attempts were much better than the first three, it seems that it could work, but it is unclear how it has to be parametrized. Try to tune with SPSA. Rescheduled after deltaLinear went to the upper limit.
15-11-06 IIv lazy_log_formula diff
LLR: 0.14 (-2.94,2.94) [0.00,5.00]
Total: 37633 W: 6084 L: 5933 D: 25616
sprt @ 10+0.1 th 7 This patch includes all theorethical ideas discussed by alwey and me, and started well on my laptop on 4 cores (133-121-277).
15-11-08 roh helper-thread-depth diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 25262 W: 4110 L: 3891 D: 17261
sprt @ 10+0.1 th 7 Increase helper thread depth when there is lots of time available.
15-11-09 sg outposts diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 15019 W: 2333 L: 2157 D: 10529
sprt @ 60+0.4 th 1 LTC. Bonus for possible outposts.
15-11-09 Ser prev_moves_3fold diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 44117 W: 8650 L: 8578 D: 26889
sprt @ 10+0.1 th 1 Improved version of http://tests.stockfishchess.org/tests/view/563fb46f0ebc59227df1710b (green). Leaves old behaviour for the root position to avoid unnecessary search cycles.
15-11-07 mco pull_clean_deepest diff
LLR: -1.47 (-2.94,2.94) [-3.00,1.00]
Total: 28644 W: 4509 L: 4629 D: 19506
sprt @ 10+0.1 th 7 Simple regression test to verify patch is correct: it is a no functional change, but has an impact only in SMP case that cannot be verified with bench, so better to test.
15-11-09 Roc Alignment diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5316 W: 996 L: 1100 D: 3220
sprt @ 10+0.1 th 1 Take 2
15-11-08 Roc RealDistance diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 41234 W: 6066 L: 6038 D: 29130
sprt @ 60+0.4 th 1 When evaluating passers, Increase king distance estimate by 1 if king cannot get any closer on next move.
15-11-07 pec not_failedlow diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 97554 W: 18807 L: 18500 D: 60247
sprt @ 10+0.1 th 1 Increment bestMoveChanges only by half if not failed low at root.
15-11-08 mbo iid_if_probcutfails diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 56391 W: 10897 L: 10772 D: 34722
sprt @ 10+0.1 th 1 Always do IID when probcut fails.