Stockfish Testing Queue

Finished - 24198 tests

30-06-15 Vo NoBonus diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 39184 W: 7379 L: 7381 D: 24424
sprt @ 15+0.05 th 1 Don't apply bonuses at depth>15. (Note the bench numbers are the same since it's higher than the bench's depth)
30-06-15 Fi smallCMH diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 36679 W: 7034 L: 6831 D: 22814
sprt @ 15+0.05 th 1 For VoyagerOne. See https://groups.google.com/forum/#!topic/fishcooking/e9kaF7kS3GI
30-06-15 lb outpost diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 18468 W: 3501 L: 3594 D: 11373
sprt @ 15+0.05 th 1 Do not cumulate Outpost with MinorBehindPawn
29-06-15 lb outpost^ diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 29566 W: 5674 L: 5727 D: 18165
sprt @ 15+0.05 th 1 Outpost from RANK_3
30-06-15 Vo CO(Fixed) diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 26446 W: 4948 L: 4984 D: 16514
sprt @ 15+0.05 th 1 Fixed Mysseno's CO patch...using file to help score captures.
30-06-15 mc raz_cond^ diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 11219 W: 2024 L: 2202 D: 6993
sprt @ 15+0.05 th 1 Test for simplification razoring condition 1 (eval + razor_margin(depth) <= alpha)...Amazingly bench number does not change!!!
29-06-15 lb outpost diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 22624 W: 4242 L: 4320 D: 14062
sprt @ 15+0.05 th 1 Outpost from RANK_5
30-06-15 lb outpost diff
LLR: -0.06 (-2.94,2.94) [0.00,4.00]
Total: 97 W: 17 L: 20 D: 60
sprt @ 15+0.05 th 1 Do not cumulate Outpost with MinorBehindPawn bonus
29-06-15 mc binary_tune2 diff
31078/40000 iterations
49522/80000 games played
80000 @ 30+0.05 th 1 Test SPSA for tuning binary variables: take 5 (every new game, range 1000, variance 40)
29-06-15 mc binary_tune2 diff
38418/40000 iterations
63000/80000 games played
80000 @ 30+0.05 th 1 Test SPSA for tuning binary variables: take 3 (every new game but very big X10 range)
29-06-15 Vo StatsLimit diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 10645 W: 2002 L: 2079 D: 6564
sprt @ 15+0.05 th 1 History Stats Limit logic tweak.
28-06-15 My KS12 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 6234 W: 1153 L: 1242 D: 3839
sprt @ 15+0.05 th 1 Hoping it's more accurate to initiate king safety based upon the enemy's piece value rather than ours.
29-06-15 My CO diff
LLR: 0.41 (-2.94,2.94) [-1.50,4.50]
Total: 210 W: 36 L: 22 D: 152
sprt @ 15+0.05 th 1 What if we consider files in capture ordering value.
29-06-15 mc binary_tune2 diff
25246/40000 iterations
43329/80000 games played
80000 @ 30+0.05 th 1 Test SPSA for tuning binary variables: take 3 (every new game but wider range)
28-06-15 lb test diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 54234 W: 10251 L: 10218 D: 33765
sprt @ 15+0.05 th 1 remove attackUnit floor
28-06-15 Vo lmrTune diff
LLR: -2.95 (-2.94,2.94) [-0.00,4.00]
Total: 45639 W: 8700 L: 8696 D: 28243
sprt @ 15+0.05 th 1 LMR Tuning Tweak
29-06-15 My KS11.1 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 22929 W: 4351 L: 4396 D: 14182
sprt @ 15+0.05 th 1 Try a variation
28-06-15 My KS11.1 diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 49918 W: 9572 L: 9544 D: 30802
sprt @ 15+0.05 th 1 Second try at king file safety. This time make sure there is significant material on board before penalising.
28-06-15 mc binary_tune diff
22260/40000 iterations
43846/80000 games played
80000 @ 30+0.05 th 1 Test SPSA for tuning binary variables: take 2 (every new move)
28-06-15 mc binary_tune^ diff
22214/40000 iterations
37518/80000 games played
80000 @ 30+0.05 th 1 Test SPSA for tuning binary variables: take 1 (every new game)
28-06-15 Ro OneWayRookV2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 16697 W: 3145 L: 3206 D: 10346
sprt @ 15+0.05 th 1 Last try
28-06-15 Ro OneWayRookV2 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 5691 W: 1039 L: 1130 D: 3522
sprt @ 15+0.05 th 1 Fix on take #1
28-06-15 Vo CaptureOrder diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 16682 W: 3119 L: 3180 D: 10383
sprt @ 15+0.05 th 1 One more shot at using mined data for sorting captures...
28-06-15 Ro OneWayRookV2 diff
LLR: -2.97 (-2.94,2.94) [0.00,6.00]
Total: 18393 W: 2869 L: 2887 D: 12637
sprt @ 60+0.05 th 1 Prevent silly Rook maneuvers @LTC SPRT[0,6]
28-06-15 lb cmh^ diff
LLR: -3.31 (-2.94,2.94) [0.00,4.00]
Total: 24555 W: 4698 L: 4788 D: 15069
sprt @ 15+0.05 th 1 cmh tweak
28-06-15 lb cmh diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 14290 W: 2670 L: 2778 D: 8842
sprt @ 15+0.05 th 1 take 2
28-06-15 lb test diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 11460 W: 2146 L: 2264 D: 7050
sprt @ 15+0.05 th 1 half attackUnit floor in king safety
26-06-15 Vo lmrMerge diff
LLR: -3.88 (-2.94,2.94) [-1.50,4.50]
Total: 99521 W: 18882 L: 18757 D: 61882
sprt @ 15+0.05 th 1 Merge decrease reduction logic.
27-06-15 Ro OneWayRookV2 diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 15348 W: 2998 L: 2851 D: 9499
sprt @ 15+0.05 th 1 Take 2
27-06-15 Vo cmhPP2 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 19975 W: 3259 L: 3137 D: 13579
sprt @ 60+0.05 th 1 Make sure we only get stats for quiet moves. See: https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/23Q---ph8zQ
27-06-15 lb eval_tune diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 8716 W: 1587 L: 1761 D: 5368
sprt @ 15+0.05 th 1 remove BishopPawns
27-06-15 Vo HistorySplit diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 11893 W: 2195 L: 2269 D: 7429
sprt @ 15+0.05 th 1 Split History at depth=5
27-06-15 lb master diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 35250 W: 6285 L: 6322 D: 22643
sprt @ 40/20 th 1 slow mover = 70. now that fishtest finally rescales increments, we can test time management patches correctly. step 3/3: tournament tc
27-06-15 Ro OneWayRookV2 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 7554 W: 1387 L: 1473 D: 4694
sprt @ 15+0.05 th 1 Prevent silly rook maneuvers.
26-06-15 Vo lmrSimple diff
LLR: -3.31 (-2.94,2.94) [-3.00,1.00]
Total: 72623 W: 13513 L: 13818 D: 45292
sprt @ 15+0.05 th 1 Remove expensive SEE function in LMR...test with recently passed LMR patch.
27-06-15 lb master diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 16612 W: 3070 L: 2858 D: 10684
sprt @ 12+0.4 th 1 slow mover = 70. now that fishtest finally rescales increments, we can test time management patches correctly. step 2/3: small base + big increment
27-06-15 lb master diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 18429 W: 3080 L: 3175 D: 12174
sprt @ 30+0.05 th 1 slow mover = 70. now that fishtest finally rescales increments, we can test time management patches correctly. step 1/3: big base + small increment
27-06-15 lb master diff
ELO: -0.00 +-59.7 (95%) LOS: 50.0%
Total: 46 W: 8 L: 8 D: 30
40000 @ 30+0.05 th 1 slow mover = 70. now that fishtest rescales increment, we can test time management patches correctly. step 1/3: big base + small increment
26-06-15 Ro CaptureCheck diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 2510 W: 416 L: 515 D: 1579
sprt @ 15+0.05 th 1 Last try. No bonus for "check + capture" on protected pawns
26-06-15 SC piece_values_see diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 6960 W: 1316 L: 1404 D: 4240
sprt @ 15+0.05 th 1 Use final tuned values from SPSA.
26-06-15 Ro CaptureCheck diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 6506 W: 1171 L: 1259 D: 4076
sprt @ 15+0.05 th 1 Moving the scoring under "Threats", and using a S(5, 5)
26-06-15 SC MVV_dynamic_OU_tuned diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 3056 W: 537 L: 635 D: 1884
sprt @ 15+0.05 th 1 Give a further try to the dynamic capture stats, this time removing the term regarding the moved piece.
24-06-15 SC see_values diff
53358/30000 iterations
75000/60000 games played
60000 @ 30+0.05 th 1 Try to use a different PieceValue for SEE-related activities. Tuning session.
24-06-15 Ro TuneCMH_Piece diff
32159/10000 iterations
61000/60000 games played
60000 @ 15+0.05 th 1 Tuning quiet move ordering by piece as suggested by VoyagerOne.
25-06-15 Vo Capture-LVA diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 83284 W: 15720 L: 15606 D: 51958
sprt @ 15+0.05 th 1 Use LVA to help sort captures.
25-06-15 Ro CaptureCheck diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 10774 W: 2014 L: 2091 D: 6669
sprt @ 15+0.05 th 1 Take 2 (6 attackUnits)
25-06-15 SC piece_values_see diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 37818 W: 7110 L: 7116 D: 23592
sprt @ 15+0.05 th 1 The spsa tuning seems to be converging to some strange values. Test whether deciding whether to use MG or EG value based on the indications of spsa tuning is good.
25-06-15 Ro CaptureCheck diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 5970 W: 1092 L: 1182 D: 3696
sprt @ 15+0.05 th 1 Add 10 attackUnits for each check which captures a piece, safely or not.
25-06-15 Ro TunedCMHPiece diff
ELO: -4.34 +-3.6 (95%) LOS: 1.0%
Total: 12816 W: 2280 L: 2440 D: 8096
30000 @ 15+0.05 th 1 Verifying the result of the first SPSA run of 60M games before tuning further.
25-06-15 lb 5ebdf2f8 diff
LLR: 2.97 (-2.94,2.94) [-5.00,0.00]
Total: 10587 W: 1691 L: 1609 D: 7287
sprt @ 60+0.05 th 1 regression test of "tune pawn shelter/storm"