Stockfish Testing Queue

Finished - 42644 tests

16-08-20 Roc PPKingPos diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10754 W: 1903 L: 1983 D: 6868
sprt @ 10+0.1 th 1 Passed pawn adjustment according to king position.
16-08-20 Roc PPKingPos2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6198 W: 1134 L: 1233 D: 3831
sprt @ 10+0.1 th 1 Take 2: According to opponent king position.
16-08-18 sg passed_pawns diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 138560 W: 26235 L: 26322 D: 86003
sprt @ 10+0.1 th 1 Simplify by using additional evaluation also for passed pawns on rank 2 and 3.
16-08-19 luc doubled_levers diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 29618 W: 5517 L: 5514 D: 18587
sprt @ 10+0.1 th 1 Lesser penalty for doubled pawns that are also levers.
16-08-19 Voy lmrCapTweak2 diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 7859 W: 1389 L: 1480 D: 4990
sprt @ 10+0.1 th 1 stc
16-08-19 tvi SE diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 76132 W: 14320 L: 14211 D: 47601
sprt @ 10+0.1 th 1 Experiment with different SE conditions when in check
16-08-19 luc cap_safety diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 9482 W: 1279 L: 1366 D: 6837
sprt @ 60+0.6 th 1 LTC (against new master) because yellow at STC and likely depth-sensitive, but low priority: Cap king safety midgame value (1? pawns)
16-08-19 luc morePawnScaling diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 18578 W: 3353 L: 3446 D: 11779
sprt @ 10+0.1 th 1 Scale factor up to 3 pawns instead of 2 (low throughput).
16-08-19 tvi SE diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 28471 W: 5223 L: 5282 D: 17966
sprt @ 10+0.1 th 1 Take 2
16-08-19 jhe promotions diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 25612 W: 4714 L: 4729 D: 16169
sprt @ 10+0.1 th 1 Do not LMR promotions
16-08-19 lan repetition_draw diff
LLR: -0.15 (-2.94,2.94) [0.00,5.00]
Total: 3327 W: 620 L: 612 D: 2095
sprt @ 10+0.1 th 1 Another take on repetition draw with larger evaluation (#7 of Stephane Nicolet)
16-08-19 Voy rpt diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 16099 W: 2891 L: 2993 D: 10215
sprt @ 10+0.1 th 1 Reduction parameter tweak
16-08-19 Voy lmrCapTweak diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 8160 W: 1440 L: 1569 D: 5151
sprt @ 10+0.1 th 1 stc
16-08-19 luc morePawnScaling diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 23654 W: 4316 L: 4340 D: 14998
sprt @ 10+0.1 th 1 My take on Stefano's patch ("Scale factor up to 4 pawns instead of 2.") with ScaleFactor not exceeding SCALE_FACTOR_NORMAL...
16-08-19 SC morePawnScaling diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 80538 W: 15049 L: 14926 D: 50563
sprt @ 10+0.1 th 1 Scale factor up to 4 pawns instead of 2.
16-08-19 luc cap_safety diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 89235 W: 16992 L: 16725 D: 55518
sprt @ 10+0.1 th 1 Cap king safety midgame value (1½ pawns)
16-08-18 lan Zobrist_side diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 19095 W: 3507 L: 3550 D: 12038
sprt @ 10+0.1 th 1 Deterministic Zobrist side key (suggestion #11 by Stephane Nicolet)
16-08-19 luc cap_safety diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 26879 W: 4947 L: 4957 D: 16975
sprt @ 10+0.1 th 1 Cap king safety midgame value (1¼ pawns)
16-08-19 Mys re2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7388 W: 1334 L: 1428 D: 4626
sprt @ 10+0.1 th 1 Last go (3rd) for now on rook endings.
16-08-19 Elb lmr_captweak diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 41531 W: 7701 L: 7714 D: 26116
sprt @ 10+0.1 th 1 LMR on captures tweak
16-08-19 tvi passed2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16270 W: 2998 L: 3053 D: 10219
sprt @ 10+0.1 th 1 Take 2
16-08-19 cib tthit_iid diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 49246 W: 9406 L: 9315 D: 30525
sprt @ 10+0.1 th 1 Lowering IID depth requirement for TT missed. Testing versus passed voyager iids patch.
16-08-19 jhe single_sided_cut diff
ELO: 0.76 +-4.2 (95%) LOS: 63.9%
Total: 10000 W: 1936 L: 1914 D: 6150
10000 @ 10+0.1 th 1 Quick measure of an idea...
16-08-19 tvi passed2 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 19306 W: 3624 L: 3666 D: 12016
sprt @ 10+0.1 th 1 Count defended squares for passed pawns
16-08-19 Elb outposts_tuned2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18897 W: 3476 L: 3520 D: 11901
sprt @ 10+0.1 th 1 Safe outposts with tuned values
16-08-19 Roc WeakColorAttacks diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 40763 W: 7570 L: 7519 D: 25674
sprt @ 10+0.1 th 1 Take 2: Only on pawns and knights
16-08-18 Elb tune_outposts2 diff
29011/30000 iterations
59530/60000 games played
60000 @ 10+0.1 th 1 Tune safe outposts (outposts that cannot be attacked by enemy minors)
16-08-19 luc cap_safety diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2963 W: 511 L: 625 D: 1827
sprt @ 10+0.1 th 1 Cap king safety midgame value (3/4 pawn)
16-08-18 cru timeman_tuned4_fix diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 3109 W: 499 L: 645 D: 1965
sprt @ 10+0.1 th 1 Changed and re-tuned move importance function. Had problems with my master branch. Hopefully fixed. Resubmitting test.
16-08-18 Roc WeakColorAttacks diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 12225 W: 2297 L: 2370 D: 7558
sprt @ 10+0.1 th 1 When opponent has exactly 1 bishop, bonus for attacking pieces on the other color.
16-08-18 tha std-threads2 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 5263 W: 793 L: 955 D: 3515
sprt @ 5+0.05 th 7 Can we get rid of thread_win32.h? Run with 7 threads a suggested by mcostalba See https://groups.google.com/d/msg/fishcooking/vuoEKP2yOCE/B7FDK5TdCwAJ
16-08-18 luc sigmoid_tropism diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6642 W: 1212 L: 1309 D: 4121
sprt @ 10+0.1 th 1 Non-linear king tropism, my (likely last) take 2.
16-08-18 jos opp_bish_threat diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 26508 W: 4993 L: 5003 D: 16512
sprt @ 10+0.1 th 1 Increase endgame threats. Take 3.
16-08-18 luc tweak_space_score diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 25763 W: 4767 L: 4835 D: 16161
sprt @ 10+0.1 th 1 Take 2: perhaps it is now fine to bump up space score?
16-08-18 jos opp_bish_threat diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 22905 W: 4241 L: 4267 D: 14397
sprt @ 10+0.1 th 1 Halve endgame score. Take 2.
16-08-18 SC double_reductions diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 9253 W: 1303 L: 1167 D: 6783
sprt @ 60+0.6 th 1 So let us try something else: maybe these double-valued reductions allow us to brutally simplify reductions init. LTC, rescheduled as simplification.
16-08-18 SC materialFile diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17318 W: 3266 L: 3316 D: 10736
sprt @ 10+0.1 th 1 Consider pawn files in material probing. How good are we after first tuning session?
16-08-18 luc sigmoid_tropism diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 14571 W: 2765 L: 2827 D: 8979
sprt @ 10+0.1 th 1 Non-linear king tropism, take 1 (don't worry, no SPSA sessions this time!)
16-08-18 Mys cs diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 16162 W: 3004 L: 3060 D: 10098
sprt @ 10+0.1 th 1 Controlled passer squares
16-08-18 Mys re1 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2873 W: 397 L: 508 D: 1968
sprt @ 10+0.1 th 1 take #2 + rocky, lucas comments
16-08-18 Mys re diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 1418 W: 182 L: 300 D: 936
sprt @ 10+0.1 th 1 Rook pawn scaling - https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/2zrTvxSdWwA the first position now roughly +0.34 not 0.8
16-08-18 Elb outposts_tuned diff
LLR: -2.94 (-2.94,2.94) [0.00,4.00]
Total: 21158 W: 3960 L: 4043 D: 13155
sprt @ 10+0.1 th 1 Use tuned outposts values
16-08-18 jos opp_bish_threat diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16129 W: 2984 L: 3040 D: 10105
sprt @ 10+0.1 th 1 Increase midgame score for threats in case of opposite bishops. (#2 on Stéphane's list)
16-08-18 SC double_reductions diff
LLR: -0.19 (-2.94,2.94) [0.00,5.00]
Total: 42 W: 1 L: 8 D: 33
sprt @ 60+0.6 th 1 So let us try something else: maybe these double-valued reductions allow us to brutally simplify reductions init. Low throughput. LTC.
16-08-17 SC materialFileTuning diff
46912/50000 iterations
100000/100000 games played
100000 @ 3+0.03 th 1 When probing material table, also consider the interactions between piece type and file of the pawn. This is based on the fact that pawn file signature only changes if also material key changes. Use a very fast tc to get some preliminary indications.
16-08-17 Elb knight_outpost diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 63625 W: 12010 L: 11944 D: 39671
sprt @ 10+0.1 th 1 Increase bonus for knight outpost (nr 9 of Stephane Nicolet's suggestions)
16-08-18 pb0 dynamic_windows diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12228 W: 2192 L: 2265 D: 7771
sprt @ 10+0.1 th 1 Take 2 (with compensation)
16-08-17 Elb tune_outposts diff
27103/30000 iterations
56024/60000 games played
60000 @ 10+0.1 th 1 Tune the outpost values as knight_outpost test indicates some improvement is possible. Lowered priority of knight_outpost test.
16-08-15 SC double_reductions diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 35598 W: 6649 L: 6553 D: 22396
sprt @ 10+0.1 th 1 So let us try something else: maybe these double-valued reductions allow us to brutally simplify reductions init. Low throughput.
16-08-18 pb0 dynamic_windows diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 21981 W: 4138 L: 4168 D: 13675
sprt @ 10+0.1 th 1 More dynamic aspiration windows in case of consecutive fail highs/lows