Stockfish Testing Queue

Finished - 35494 tests

16-09-30 atu rook_mobility_bonus diff
29264/30000 iterations
60000/60000 games played
60000 @ 20+0.2 th 1 Replace the formula for a rook trapped by the king by an const array and tune its values. There are also some related changes.
16-10-01 aji fpt2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5049 W: 884 L: 988 D: 3177
sprt @ 10+0.1 th 1 If eval comes from TT, relax the fulity pruning check depending on tte->depth() : STC
16-09-30 sni safe_checks3 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 50474 W: 9033 L: 8962 D: 32479
sprt @ 10+0.1 th 1 Simplify SafeCheck
16-09-30 sni pinners diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 34011 W: 6069 L: 6051 D: 21891
sprt @ 10+0.1 th 1 Don't apply the penalty for our pinned queen if we attack back the pinners.
16-10-01 Roc TropismCF diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 1990 W: 312 L: 464 D: 1214
sprt @ 10+0.1 th 1 Small adjustment to tropism area when king is on c or f file.
16-09-30 sni combo4 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 32005 W: 5709 L: 5757 D: 20539
sprt @ 10+0.1 th 1 Combo of two yellows (and in fact a slight simplification)
16-09-30 fau tune diff
9865/10000 iterations
20000/20000 games played
20000 @ 10+0.1 th 1 Fastly tuning some untuned values
16-09-30 aji fpt diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 26983 W: 4805 L: 4816 D: 17362
sprt @ 10+0.1 th 1 futility prune tweak : STC
16-09-30 SC stonewall3 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 3364 W: 557 L: 721 D: 2086
sprt @ 10+0.1 th 1 Retire unsupported penalty altogether.
16-09-30 Voy inCheck diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 9640 W: 1786 L: 1621 D: 6233
sprt @ 10+0.1 th 1 Can we SEE prune inCheck?
16-09-30 mbo odd_qmob diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 13744 W: 2345 L: 2455 D: 8944
sprt @ 10+0.1 th 1 Reduce bonus for a too mobile queen. Lower throughput.
16-09-30 aji fptuned2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 31672 W: 5637 L: 5629 D: 20406
sprt @ 10+0.1 th 1 Tweak one of the tuned values(maybe I interpreted rounding incorrectly): STC
16-09-30 luc pawn_psqt diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 56588 W: 10176 L: 10140 D: 36272
sprt @ 10+0.1 th 1 Only slight modifications, for values I found a bit strange...
16-09-30 Voy qsSEE diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7610 W: 1313 L: 1406 D: 4891
sprt @ 10+0.1 th 1 stc
16-09-28 luc same_color_supported_tu diff
19686/20000 iterations
40000/40000 games played
40000 @ 20+0.2 th 1 It is said the the worst pawns for the bishop on same color square are the central pawns... Low throughput
16-09-30 pb0 historyFormula4 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 11858 W: 2101 L: 2176 D: 7581
sprt @ 10+0.1 th 1 Take 6: use 2 different formulas, depending on rootDepth is low/high
16-09-29 sni octopus diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 99319 W: 17655 L: 17478 D: 64186
sprt @ 10+0.1 th 1 Take 3, change = -15
16-09-30 Elb bad_bishop3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 1991 W: 309 L: 426 D: 1256
sprt @ 10+0.1 th 1 Now only consider inactive bad bishops
16-09-29 sni octopus diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 49731 W: 8897 L: 8812 D: 32022
sprt @ 10+0.1 th 1 Try the opposite
16-09-30 pb0 dangerous_checks diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13891 W: 2474 L: 2540 D: 8877
sprt @ 10+0.1 th 1 extend dangerous checks despite movecount
16-09-30 Elb bad_bishop3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9766 W: 1714 L: 1798 D: 6254
sprt @ 10+0.1 th 1 Bugfixed version
16-09-30 Mys ev diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7517 W: 1325 L: 1418 D: 4774
sprt @ 10+0.1 th 1 Pieces in proximity to our king imposed by the enemy reduces king safety.
16-09-30 pb0 historyFormula4 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 3224 W: 528 L: 640 D: 2056
sprt @ 10+0.1 th 1 exponent = 2.184, linear coefficient = 10 and doubling ceiling & multiplier values in movepick.h so that depth 17 keeps to be the highest one to update the history.
16-09-30 luc simple_psqt diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 17970 W: 3099 L: 3195 D: 11676
sprt @ 10+0.1 th 1 I generated from scratch some new eg psqt values for pawns not too different from the current ones (RMSD = 2.89), and a local test scored very well, so I'd better try those out too...
16-09-29 Voy sqss diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 109365 W: 19231 L: 19571 D: 70563
sprt @ 10+0.1 th 1 Simplify SEE in qsearch
16-09-30 aji fptuned diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8370 W: 1486 L: 1576 D: 5308
sprt @ 10+0.1 th 1 Add staticEval check for futility pruning(use tuned values); STC
16-09-29 eel Michel's_razorcode diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 76209 W: 13515 L: 13416 D: 49278
sprt @ 10+0.1 th 1 As parameterchange because not sure code should have been made inactive
16-09-27 fau tune diff
38938/40000 iterations
80000/80000 games played
80000 @ 20+0.2 th 1 Tuning Rook's Values and some Rook's bonuses altogether.
16-09-30 Elb bad_bishop3 diff
LLR: -0.00 (-2.94,2.94) [0.00,5.00]
Total: 31 W: 6 L: 6 D: 19
sprt @ 10+0.1 th 1 Take 2: hopefully now with fixed bench
16-09-29 aji tune_fpcheck diff
19727/20000 iterations
40000/40000 games played
40000 @ 10+0.1 th 1 Tune futility prune static eval check
16-09-30 Voy lmrIdea diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 16156 W: 2774 L: 2832 D: 10550
sprt @ 10+0.1 th 1 See how this idea goes...
16-09-30 cib reducedDepth diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12207 W: 2088 L: 2162 D: 7957
sprt @ 10+0.1 th 1 Extra penalty for refuted quiet move that pass the reduced depth search.
16-09-30 luc simple_psqt diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 8275 W: 1399 L: 1527 D: 5349
sprt @ 10+0.1 th 1 Tuned endgame values for passed pawn rank bonuses
16-09-29 aji futility_pruning diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 63897 W: 11505 L: 11360 D: 41032
sprt @ 10+0.1 th 1 Add a staticEval check for futility pruning: STC
16-09-28 luc simple_psqt_tune_passed diff
18334/20000 iterations
37264/40000 games played
40000 @ 20+0.2 th 1 Endgame Passed values were never really tuned... But I will stop in case they go nowhere!
16-09-28 atu rook_mobility_bonus diff
37252/40000 iterations
80000/80000 games played
80000 @ 20+0.2 th 1 Tune MobilityBonus[ROOK][i] for i = 0, 1, 2, 3 (fixed braches)
16-09-29 SC stonewall2 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 29845 W: 5287 L: 5287 D: 19271
sprt @ 10+0.1 th 1 I hope I got everything right the end. Thanks to Rocky...
16-09-29 Voy sqss2 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 8494 W: 1452 L: 1623 D: 5419
sprt @ 10+0.1 th 1 Even bigger simplification...
16-09-29 aji futility_pruning2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 15317 W: 2667 L: 2728 D: 9922
sprt @ 10+0.1 th 1 Take 2: STC
16-09-29 aji probcut_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13155 W: 2278 L: 2348 D: 8529
sprt @ 10+0.1 th 1 probcut tweak STC
16-09-29 Voy trf diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 11203 W: 1930 L: 2048 D: 7225
sprt @ 10+0.1 th 1 Try 2.08
16-09-29 Mys ea diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5170 W: 894 L: 997 D: 3279
sprt @ 10+0.1 th 1 Stationed pieces in king safety
16-09-29 sni nimzovitch'' diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 21850 W: 3889 L: 3922 D: 14039
sprt @ 10+0.1 th 1 Nimzovitch: bonus for protected reachable queen outposts
16-09-29 sni quality diff
LLR: -1.63 (-2.94,2.94) [0.00,5.00]
Total: 9595 W: 1737 L: 1765 D: 6093
sprt @ 10+0.1 th 1 Avoid losing quality: exclude squares attacked by knight or bishop from rook mobility
16-09-29 sni bishop_pair diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8997 W: 1627 L: 1714 D: 5656
sprt @ 10+0.1 th 1 Try to keep more bishop pairs
16-09-29 Elb bad_bishop diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5490 W: 920 L: 1022 D: 3548
sprt @ 10+0.1 th 1 Only give BishopPawn penalty if bishop is behind front most pawn.
16-09-29 Voy trf diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 75753 W: 13421 L: 13323 D: 49009
sprt @ 10+0.1 th 1 Take 2
16-09-28 kan opp_bish_4 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 74740 W: 13329 L: 13140 D: 48271
sprt @ 10+0.1 th 1 Test Rocky's suggestion.
16-09-29 Voy killer diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16510 W: 2888 L: 2944 D: 10678
sprt @ 10+0.1 th 1 If main killer doesn't exist...then use prior main killer.
16-09-28 sni octopus diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 13259 W: 2282 L: 2393 D: 8584
sprt @ 10+0.1 th 1 Octopus?