Stockfish Testing Queue

Pending - 2 tests 0.1 hrs

28-04-17 II history diff
LLR: 0.88 (-2.94,2.94) [0.00,4.00]
Total: 63000 W: 11487 L: 11226 D: 40287
sprt @ 10+0.1 th 1 Linear history stats.
29-04-17 II linear_stats diff
LLR: 0.69 (-2.94,2.94) [0.00,4.00]
Total: 57000 W: 10302 L: 10073 D: 36625
sprt @ 10+0.1 th 1 Take 3 - without excluding low depths.

Active - 1 tests

30-04-17 II king_safety diff
LLR: 1.62 (-2.94,2.94) [0.00,4.00]
Total: 11477 W: 2137 L: 2012 D: 7328
sprt @ 10+0.1 th 1 Combo with rook mobility bonus.

Finished - 312 tests

30-04-17 II king_safety diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 29142 W: 5189 L: 5247 D: 18706
sprt @ 10+0.1 th 1 King safety parameters 'by feeling' - based on several tunings.
28-04-17 II history diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5638 W: 948 L: 1049 D: 3641
sprt @ 10+0.1 th 1 Don't update history stats at low depths.
25-04-17 II king_safety diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 119975 W: 21533 L: 21284 D: 77158
sprt @ 10+0.1 th 1 King safety - take 3 (tuned values).
25-04-17 II tune_stats diff
27681/110000 iterations
56207/220000 games played
220000 @ 10+0.1 th 1 Continue tuning history stats with better ck values.
22-04-17 II tune_king diff
19341/20000 iterations
39114/40000 games played
40000 @ 30+0.3 th 1 Further tune king safety. Throughput 300 because of the used TC.
22-04-17 II tune_stats diff
38402/150000 iterations
78987/300000 games played
300000 @ 10+0.1 th 1 Trying to reveal true structure of history stats. Throughput 300 because of the number of games used. I cut updates at depth 12 because it is appropriate so for STC, and plan to extend to higher depths by analogy.
21-04-17 II king_safety diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 52546 W: 7131 L: 6844 D: 38571
sprt @ 60+0.6 th 1 LTC (I think that 130K games is enough to reduce priority of the first test and start another) - Take 2 (values obtained by recent local tuning)
19-04-17 II king_safety diff
LLR: 1.24 (-2.94,2.94) [0.00,4.00]
Total: 136952 W: 18146 L: 17725 D: 101081
sprt @ 60+0.6 th 1 LTC: King safety - tuned values.
20-04-17 II king_safety diff
LLR: 2.97 (-2.94,2.94) [0.00,4.00]
Total: 58648 W: 10883 L: 10524 D: 37241
sprt @ 10+0.1 th 1 Take 2 (values obtained by recent local tuning)
19-04-17 II king_safety diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 16084 W: 2998 L: 2787 D: 10299
sprt @ 10+0.1 th 1 King safety - tuned values.
17-04-17 II tune_eval diff
38390/40000 iterations
78269/80000 games played
80000 @ 20+0.2 th 1 Trying to tune king safety. TC 20+0.2 is by purpose. Local tuning at 40+0.4 will be used as a 'second opinion'.
17-04-17 II history_stats diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 4251 W: 701 L: 808 D: 2742
sprt @ 10+0.1 th 1 Trying to avoid floating point arithmetic, as suggested by vondele.
16-04-17 II nullmove diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 2798 W: 470 L: 584 D: 1744
sprt @ 10+0.1 th 1 Take 2. Increasing reductions by 1.
16-04-17 II nullmove diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4416 W: 738 L: 844 D: 2834
sprt @ 10+0.1 th 1 Null move reductions - tuned values.
16-04-17 II history_stats diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 4692 W: 808 L: 914 D: 2970
sprt @ 10+0.1 th 1 Take 3. If there will be no significant progress, I give up with this idea.
13-04-17 II history_stats diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 4583 W: 757 L: 863 D: 2963
sprt @ 10+0.1 th 1 History stats inspired by tuning. Take 2.
13-04-17 II history_stats diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2904 W: 465 L: 578 D: 1861
sprt @ 10+0.1 th 1 History stats inspired by tuning. Take 1.
08-04-17 II tune_rhist diff
77637/80000 iterations
160000/160000 games played
160000 @ 10+0.1 th 1 Trying to tune history and nullmove reductions from scratch.
01-04-17 II delta diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 50821 W: 8997 L: 9239 D: 32585
sprt @ 10+0.1 th 1 I think that I don't want to force delta patch (Elo gain seems to be marginal), but I'll take two attempts trying to remove lazy evaluation with delta patch - Take 1.
01-04-17 II rook_mobility diff
LLR: -4.59 (-2.94,2.94) [0.00,4.00]
Total: 42472 W: 7569 L: 7668 D: 27235
sprt @ 10+0.1 th 1 Rook and Queen MobilityBonus - tuned values.
31-03-17 II tune_eval diff
37572/40000 iterations
78769/80000 games played
80000 @ 10+0.1 th 1 Tune QueenMobilityBonus.
31-03-17 II delta diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 44781 W: 8173 L: 8107 D: 28501
sprt @ 10+0.1 th 1 Non-constant delta, take 3.
31-03-17 II delta diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 27792 W: 5013 L: 5020 D: 17759
sprt @ 10+0.1 th 1 Non-constant delta, take 2.
31-03-17 II delta diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 21318 W: 3743 L: 3779 D: 13796
sprt @ 10+0.1 th 1 Last idea on non-constant delta.
28-03-17 II pawn_seed diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 25868 W: 4532 L: 4601 D: 16735
sprt @ 10+0.1 th 1 Last try. It seems this is already well tuned.
27-03-17 II tune_seed diff
32646/40000 iterations
67057/80000 games played
80000 @ 10+0.1 th 1 Continue tuning Pawn Seed Array.
25-03-17 II rook_mobility diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 27386 W: 4961 L: 5024 D: 17401
sprt @ 10+0.1 th 1 RookMobility after further tuning.
23-03-17 II rook_mobility diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 77998 W: 10097 L: 10042 D: 57859
sprt @ 60+0.6 th 1 LTC. Rook mobility. Tuned values.
23-03-17 II pawn_seed diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 67689 W: 12068 L: 11996 D: 43625
sprt @ 10+0.1 th 1 Take 2. Some of tuned values (fixed).
23-03-17 II pawn_seed diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 14153 W: 2418 L: 2526 D: 9209
sprt @ 10+0.1 th 1 Pawn seed parameter tweak. Tuned values.
19-03-17 II rook_mobility diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 211425 W: 38262 L: 37391 D: 135772
sprt @ 10+0.1 th 1 Rook mobility. Tuned values.
22-03-17 II tune_seed diff
35375/40000 iterations
72108/80000 games played
80000 @ 10+0.1 th 1 Tune pawn seed with pretty high ck values and Rk=0.0005.
20-03-17 II tune_eval diff
19773/20000 iterations
40000/40000 games played
40000 @ 10+0.1 th 1 Continue tuning RookMobility, now with ck=30, Rk=0.0005.
17-03-17 II tune_eval diff
37433/40000 iterations
78012/80000 games played
80000 @ 10+0.1 th 1 Tune evaluation #3. As the First two tunings were neutral try to use higher ck values, but compensate with lower Rk (ck=20, Rk=0.0005).
18-03-17 II delta diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 26155 W: 4531 L: 4600 D: 17024
sprt @ 10+0.1 th 1 Can we increase delta further?
15-03-17 II bishop_mobility diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 32220 W: 5606 L: 5654 D: 20960
sprt @ 10+0.1 th 1 Bishop MobilityBonus. Tuning patch was stopped by one worker, but let's check values after 52K games.
13-03-17 II tune_eval diff
25908/40000 iterations
52690/80000 games played
80000 @ 10+0.1 th 1 Tune evaluation #2. 20% lower ck values this time.
13-03-17 II knight_mobility diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 44827 W: 7872 L: 7878 D: 29077
sprt @ 10+0.1 th 1 Knight mobility bonus. Tuned values.
12-03-17 II tune_eval diff
36912/40000 iterations
79526/80000 games played
80000 @ 10+0.1 th 1 Tune evaluation #1. I'll try to retune the whole evaluation function, tuning 8-32 parameters in one session. Throughput 400.
11-03-17 II outpost diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 25668 W: 4567 L: 4636 D: 16465
sprt @ 10+0.1 th 1 Tuned outpost values.
11-03-17 II nmp diff
LLR: -3.18 (-2.94,2.94) [0.00,5.00]
Total: 9097 W: 1562 L: 1658 D: 5877
sprt @ 10+0.1 th 1 NullMove Search. Tuned values. Last try.
10-03-17 II tune_nmp diff
29150/30000 iterations
59035/60000 games played
60000 @ 10+0.1 th 1 Trying to tune NullMove search, following ElbertoOne's idea.
04-03-17 II tune_outpost diff
58622/60000 iterations
119450/120000 games played
120000 @ 10+0.1 th 1 Fine tuning outpost values, checking theory obtained from the built-in simulator.
10-03-17 II nmp diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 14915 W: 2627 L: 2689 D: 9599
sprt @ 10+0.1 th 1 NullMove tweak; testing for good starting point for tuning.
10-03-17 II tune_nmp diff
475/20000 iterations
994/40000 games played
40000 @ 10+0.1 th 1 Trying to tune NullMove search, following ElbertoOne's idea.
29-01-17 II update diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 4538 W: 747 L: 912 D: 2879
sprt @ 10+0.1 th 1 Stats simplification, take 2.
26-01-17 II tmm_simple diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 6025 W: 995 L: 1161 D: 3869
sprt @ 10+0.1 th 1 SCSPSA Test #2 - time management simplification.
26-01-17 II psqt diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 21577 W: 3891 L: 3974 D: 13712
sprt @ 10+0.1 th 1 SCSPSA Test #1 - king PSQT.
22-01-17 II update diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 3511 W: 548 L: 710 D: 2253
sprt @ 10+0.1 th 1 Stats simplification (joint idea by Stefan Geschwentner and Ivan Ivec). First untuned guess.
18-01-17 II update diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 3543 W: 596 L: 706 D: 2241
sprt @ 10+0.1 th 1 SF statistics, take 1.