Stockfish Testing Queue

Finished - 49271 tests

15-11-23 SC tuneRazor diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 3240 W: 541 L: 688 D: 2011
sprt @ 10+0.1 th 1 Have a go at razoring at higher depths while there.
15-11-23 aji fmc_tuned diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 4310 W: 786 L: 930 D: 2594
sprt @ 10+0.1 th 1 Tuned FutilityMoveCount values : STC
15-11-23 aji fmc_tuned diff
LLR: -1.00 (-2.94,2.94) [0.00,4.00]
Total: 714 W: 108 L: 159 D: 447
sprt @ 20+0.2 th 1 Tuned FutilityMoveCount values : STC Using 20+0.2 since it is required to cover depths of upto 16. Also it was tuned at 20+0.2.
15-11-23 Voy hppv diff
LLR: -1.57 (-2.94,2.94) [0.00,5.00]
Total: 2811 W: 520 L: 575 D: 1716
sprt @ 10+0.1 th 1 Don't history-prune pv nodes. Fix bench.
15-11-23 pec tm diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 32622 W: 6321 L: 6050 D: 20251
sprt @ 10+0.1 th 1 Try mbootsector's slow mover 84. It passed for similar patch, and in the context of previous patch it will redistribute more time to the beginning of game, which probably should be beneficial for incremental tc at the expense of sudden death tc
15-11-21 IIv NRF_tune diff
39762/40000 iterations
80000/80000 games played
80000 @ 20+0.2 th 1 Tuning new reduction formula. I'm also running my local tuning method to achieve maximum out of this. My reasons are explained on the forum: https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/eSqiHjdMoXY (sorry for a mistake first run)
15-11-23 pec tm diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10070 W: 1389 L: 1473 D: 7208
sprt @ 60+0.4 th 1 LTC. Use previous search score to soft-stop current search earlier
15-11-22 pec tm diff
ELO: 7.23 +-3.0 (95%) LOS: 100.0%
Total: 20000 W: 4161 L: 3745 D: 12094
20000 @ 15 th 1 Measure elo at sudden death tc. Use previous search score to soft-stop current search earlier
15-11-23 Voy CME4 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 43526 W: 8347 L: 8280 D: 26899
sprt @ 10+0.1 th 1 Take 4: One more shot at this...
15-11-21 mco deepest_no_score diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 6868 W: 1122 L: 1289 D: 4457
sprt @ 5+0.1 th 7 Try to give a sense to deepest thread best move
15-11-22 pec tm diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 44959 W: 8681 L: 8356 D: 27922
sprt @ 10+0.1 th 1 Use previous search score to soft-stop current search earlier
15-11-22 Voy CME3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 6759 W: 1231 L: 1328 D: 4200
sprt @ 10+0.1 th 1 Take 3.
15-11-22 sni threat_weight diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 30467 W: 5826 L: 5817 D: 18824
sprt @ 10+0.1 th 1 Threats weight={300, 240}
15-11-22 Roc OverAttack diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7372 W: 1376 L: 1470 D: 4526
sprt @ 10+0.1 th 1 Take 1
15-11-22 SC varianceRazor diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 4729 W: 838 L: 944 D: 2947
sprt @ 10+0.1 th 1 Last attempt: values tuned locally.
15-11-22 Voy FHt diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 18411 W: 3472 L: 3565 D: 11374
sprt @ 10+0.1 th 1 FH bonus tweak
15-11-21 mco deepest_no_score diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 3406 W: 534 L: 696 D: 2176
sprt @ 5+0.1 th 7 Try to give a sense to deepest thread best move: take 2
15-11-21 aji tune_fmc diff
25361/40000 iterations
51634/80000 games played
80000 @ 20+0.2 th 1 Try to create futilitymovecount using tuned values instead of formula.
15-11-22 Voy CME2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 83092 W: 16130 L: 15885 D: 51077
sprt @ 10+0.1 th 1 Take 2: Don't score evasions in QS.
15-11-22 sni pieces2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2763 W: 476 L: 591 D: 1696
sprt @ 10+0.1 th 1 More pawn asymmetry when attacking
15-11-22 lbr equal_smp diff
LLR: -2.95 (-2.94,2.94) [-4.00,0.00]
Total: 3086 W: 555 L: 728 D: 1803
sprt @ 5+0.05 th 1 let's see if we regress on 1 thread, before doing any SMP test
15-11-21 jos aspiration_tweak2 diff
LLR: -3.20 (-2.94,2.94) [0.00,4.00]
Total: 45935 W: 8821 L: 8829 D: 28285
sprt @ 10+0.1 th 1 Open aspiration window depending on the current score. (Fixed version. Thanks Stefan!)
15-11-22 sg passed_pawns2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10510 W: 1991 L: 2071 D: 6448
sprt @ 10+0.1 th 1 Tweak endgame value of passed pawns
15-11-22 sni pieces2 diff
LLR: -1.15 (-2.94,2.94) [0.00,5.00]
Total: 1068 W: 196 L: 241 D: 631
sprt @ 10+0.1 th 1 Keep many pieces when attacking
15-11-18 luc addDepthByBF diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 58447 W: 10202 L: 10085 D: 38160
sprt @ 5+0.1 th 7 Set depth for helper threads based on the branching factor, i.e. higher when we've been iterating with less nodes. Guess values. Comments welcome!
15-11-21 sni combo2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 27599 W: 5351 L: 5354 D: 16894
sprt @ 10+0.1 th 1 Yellows together
15-11-21 Voy CME diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 23112 W: 4444 L: 4468 D: 14200
sprt @ 10+0.1 th 1 Use cmh to score evasions. Thanks SG for helping debug the bench# issue.
15-11-21 IIv NRF_tune diff
6/40000 iterations
10/80000 games played
80000 @ 20+0.2 th 1 Tuning new reduction formula. I'm also running my local tuning method to achieve maximum out of this. My reasons are explained on the forum: https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/eSqiHjdMoXY
15-11-18 gog master diff
ELO: 63.30 +-1.9 (95%) LOS: 100.0%
Total: 40000 W: 10289 L: 3081 D: 26630
40000 @ 60+0.05 th 1
15-11-21 jos aspiration_tweak2 diff
LLR: -0.31 (-2.94,2.94) [0.00,4.00]
Total: 1319 W: 252 L: 264 D: 803
sprt @ 10+0.1 th 1 Open aspiration window depending on the current score.
15-11-21 sni supported_lever5 diff
LLR: -4.27 (-2.94,2.94) [0.00,5.00]
Total: 32550 W: 4711 L: 4768 D: 23071
sprt @ 60+0.4 th 1 LTC: Supported levers with higher values in midgame than in endgame
15-11-20 mbo tm_resolve_faillow diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 75056 W: 11664 L: 11263 D: 52129
sprt @ 40+0.4 th 1 LTC. Time management: Use the score from previous search to resolve fail lows. Slow Mover=84.
15-11-20 Voy bt diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 29795 W: 5719 L: 5713 D: 18363
sprt @ 10+0.1 th 1 refutation bonus tweak
15-11-20 sni supported_lever5 diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 19036 W: 3759 L: 3547 D: 11730
sprt @ 10+0.1 th 1 Supported levers with higher values in midgame than in endgame
15-11-20 Voy spk diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16417 W: 3141 L: 3195 D: 10081
sprt @ 10+0.1 th 1 Don't prune main killer.
15-11-20 roh Choose-Best-Thread diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6422 W: 1171 L: 1269 D: 3982
sprt @ 10+0.1 th 1 Repeating search at same depth seems to add little extra time spent but gives better search results.
15-11-20 sg rook_outposts diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12428 W: 2384 L: 2455 D: 7589
sprt @ 10+0.1 th 1 Try rook outposts
15-11-17 Fis spinLocks diff
LLR: -2.39 (-2.94,2.94) [-3.00,1.00]
Total: 128000 W: 21800 L: 22135 D: 84065
sprt @ 10+0.1 th 3 Remove windows specific stuff and always use spinlocks implemented using standard STL.
15-11-19 sni initiative2 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 79751 W: 15451 L: 15324 D: 48976
sprt @ 10+0.1 th 1 Raise initiative() value by 25%
15-11-20 mbo tm_resolve_faillow diff
LLR: 2.94 (-2.94,2.94) [0.00,5.00]
Total: 14495 W: 2903 L: 2711 D: 8881
sprt @ 10+0.1 th 1 Time management: Use the score from previous search to resolve fail lows. Slow Mover=84.
15-11-19 sni threat_weight diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 61755 W: 12071 L: 11920 D: 37764
sprt @ 10+0.1 th 1 Raise threats weight by 17%
15-11-19 IIv NRF diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 34745 W: 6616 L: 6589 D: 21540
sprt @ 10+0.1 th 1 New reduction formula - the first intuitive try.
15-11-19 sni king_safety4 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7434 W: 1395 L: 1489 D: 4550
sprt @ 10+0.1 th 1 Slightly negative king safety in endgame
15-11-19 Roc MinorPawnCoord diff
ELO: -3.35 +-5.0 (95%) LOS: 9.6%
Total: 7160 W: 1363 L: 1432 D: 4365
20000 @ 10+0.1 th 1 MinorBehindPawn is a feature worth about 20 ELO. Let's see how much we loose (or gain !) by using a different concept.
15-11-19 sni pawn_weight diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 69312 W: 13500 L: 13410 D: 42402
sprt @ 10+0.1 th 1 Pawn weight tweak
15-11-17 Voy rf diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 102415 W: 19651 L: 19445 D: 63319
sprt @ 10+0.1 th 1 Try a much higher weight for depth, in reduction formula.
15-11-19 bin mpv_thread diff
LLR: 0.20 (-2.94,2.94) [0.00,5.00]
Total: 3000 W: 550 L: 529 D: 1921
sprt @ 5+0.1 th 7 Assign thread 1 to 2 PVs
15-11-19 bin thread_BF diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 3593 W: 555 L: 664 D: 2374
sprt @ 5+0.1 th 7 Thread based pruning/reductions/SE depth
15-11-19 sni isolated_h diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 37979 W: 7224 L: 7248 D: 23507
sprt @ 10+0.1 th 1 Half value
15-11-18 mbo tm_resolve_faillow diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 57444 W: 8367 L: 8282 D: 40795
sprt @ 60+0.4 th 1 LTC. Time management: Use the score from previous search to resolve fail lows.