Stockfish Testing Queue

Finished - 19396 tests

14-08-13 ra master diff
ELO: 6.81 +-75.4 (95%) LOS: 57.1%
Total: 51 W: 16 L: 15 D: 20
1000 @ 15+0.05 th 1 Quick test for crashes.
13-08-13 pe decrease_improving diff
LLR: -2.95 (-2.94,2.94)
Total: 16045 W: 2906 L: 2928 D: 10211
sprt @ 60+0.05 th 1 Long TC: Decrease pruning if eval is improving. Take 2
13-08-13 jh easymove diff
ELO: -2.10 +-2.9 (95%) LOS: 7.6%
Total: 20000 W: 3499 L: 3620 D: 12881
20000 @ 60+0.05 th 1 Determine value of current easy move implementation.
13-08-13 pe decrease_improving diff
LLR: 2.95 (-2.94,2.94)
Total: 18613 W: 3907 L: 3747 D: 10959
sprt @ 15+0.05 th 1 Decrease pruning if eval is improving. Take 2
13-08-13 ra TT_Store_A diff
ELO: 0.19 +-3.1 (95%) LOS: 54.9%
Total: 20000 W: 4067 L: 4056 D: 11877
20000 @ 15+0.05 th 1 Tweak TT replacement. A little simplification.
13-08-13 mc lmr diff
LLR: -2.95 (-2.94,2.94)
Total: 22140 W: 4562 L: 4607 D: 12971
sprt @ 15+0.05 th 1 Reduce less when improving: take 3
13-08-13 mc lmr diff
LLR: -2.96 (-2.94,2.94)
Total: 8915 W: 1830 L: 1913 D: 5172
sprt @ 15+0.05 th 1 Search deeper along variations that are improving: take 2
06-08-13 tv counter_cut diff
LLR: -2.97 (-2.94,2.94)
Total: 26454 W: 4782 L: 4752 D: 16920
sprt @ 60+0.05 th 1 Long TC for Gary: Reduce countermoves less in cut nodes
13-08-13 mc lmr diff
LLR: -2.97 (-2.94,2.94)
Total: 29934 W: 6280 L: 6303 D: 17351
sprt @ 15+0.05 th 1 Search deeper along variations that are improving: take 1
13-08-13 tv PiecePawn diff
LLR: -2.96 (-2.94,2.94)
Total: 10050 W: 2156 L: 2236 D: 5658
sprt @ 15+0.05 th 1 No space, scale piece-pawn bonus with rank
12-08-13 tv PiecePawn diff
LLR: 2.96 (-2.94,2.94)
Total: 10218 W: 1932 L: 1775 D: 6511
sprt @ 60+0.05 th 1 Ltc: Any pawn in front of minor gives bonus as suggested by Marco
12-08-13 tv LongBishop diff
LLR: -2.21 (-2.94,2.94)
Total: 2846 W: 517 L: 581 D: 1748
sprt @ 60+0.05 th 1 Ltc: Remove center pawn restriction
12-08-13 tv PiecePawn diff
LLR: -1.67 (-2.94,2.94)
Total: 4378 W: 892 L: 966 D: 2520
sprt @ 15+0.05 th 1 No space and our pawn in front of minor (SPRT -4.5/1.5)
04-08-13 tv LongBishop diff
LLR: -1.77 (-2.94,2.94)
Total: 22197 W: 4707 L: 4708 D: 12782
sprt @ 15+0.05 th 1 Pure defense fianchetto
10-08-13 tv master diff
ELO: 51.20 +-3.2 (95%) LOS: 100.0%
Total: 20000 W: 5953 L: 3027 D: 11020
20000 @ 60+0.05 th 1 Regression test of latest "space" version
04-08-13 tv LongBishop diff
LLR: 2.96 (-2.94,2.94)
Total: 22768 W: 4951 L: 4776 D: 13041
sprt @ 15+0.05 th 1 Remove center pawn restriction
03-08-13 jk fianchetto_bishop diff
LLR: 2.96 (-2.94,2.94)
Total: 105752 W: 22388 L: 21977 D: 61387
sprt @ 15+0.05 th 1 Give a bonus for fianchettoed bishop protecting the king
03-08-13 pe evaltune diff
LLR: -2.97 (-2.94,2.94)
Total: 38396 W: 7119 L: 7025 D: 24252
sprt @ 60+0.05 th 1 Long TC. Combo of two patches tested before and having high probability of being positive. Looks like it gets around 2 elo from TC doubling , and it was at around 50% at 15+0.05
11-08-13 tv EvalRound diff
LLR: -2.97 (-2.94,2.94)
Total: 4760 W: 946 L: 1041 D: 2773
sprt @ 15+0.05 th 1 Round away from zero so that 1 - 3 round to 4 and -1 - -3 round to -4 instead of 0
11-08-13 tv NoSpace diff
LLR: -2.95 (-2.94,2.94)
Total: 1383 W: 324 L: 447 D: 612
sprt @ 15+0.05 th 1 Test a clean no space SF (SPRT -4.5/1.5)
11-08-13 mc master diff
ELO: 2.35 +-3.1 (95%) LOS: 93.2%
Total: 20000 W: 4189 L: 4054 D: 11757
20000 @ 15+0.05 th 1 Verify the fix on grain size does not regress
11-08-13 tv EvalRound diff
LLR: 0.16 (-2.94,2.94)
Total: 379 W: 85 L: 78 D: 216
sprt @ 15+0.05 th 1 Other round implementation (round away from zero so that -1 and 1 no longer round to 0)
10-08-13 tv PiecePawn diff
LLR: -2.95 (-2.94,2.94)
Total: 17158 W: 3175 L: 3326 D: 10657
sprt @ 60+0.05 th 1 Test LTC (sprt -4.5/1.5) because it is a big simplification which scores ~50% at STC: Without space, any pawn and including first rank.
10-08-13 tv PiecePawn diff
LLR: -0.42 (-2.94,2.94)
Total: 28234 W: 6027 L: 5961 D: 16246
sprt @ 15+0.05 th 1 Without space, any pawn and including first rank
10-08-13 tv CheckNotImprove diff
LLR: -1.14 (-2.94,2.94)
Total: 12910 W: 2655 L: 2660 D: 7595
sprt @ 15+0.05 th 1 Test old non improving after check condition versus improving after check
09-08-13 tv PiecePawn diff
LLR: -0.19 (-2.94,2.94)
Total: 26000 W: 4844 L: 4716 D: 16440
sprt @ 60+0.05 th 1 Ltc: Any pawn in front of minor gives bonus as suggested by Marco
04-08-13 tv PiecePawn diff
LLR: 2.97 (-2.94,2.94)
Total: 7629 W: 1645 L: 1515 D: 4469
sprt @ 15+0.05 th 1 Any pawn in front of minor gives bonus as suggested by Marco
09-08-13 tv Time diff
LLR: -2.16 (-2.94,2.94)
Total: 5904 W: 1050 L: 1096 D: 3758
sprt @ 60+0.05 th 1 LTC: Adjust 20%
09-08-13 tv Time diff
LLR: -1.79 (-2.94,2.94)
Total: 4725 W: 940 L: 992 D: 2793
sprt @ 15+0.05 th 1 Code cleanup and 33% of deviation
09-08-13 tv Time diff
LLR: 0.40 (-2.94,2.94)
Total: 4951 W: 1055 L: 1026 D: 2870
sprt @ 15+0.05 th 1 Use 50% of the ideal time deviation of the previous move
09-08-13 tv Time diff
LLR: 2.94 (-2.94,2.94)
Total: 7133 W: 1516 L: 1389 D: 4228
sprt @ 15+0.05 th 1 Adjust 20%
09-08-13 tv Time diff
LLR: -2.95 (-2.94,2.94)
Total: 4637 W: 935 L: 1030 D: 2672
sprt @ 15+0.05 th 1 Try to compensate for time used in previous move
08-08-13 mc search diff
LLR: -1.45 (-2.94,2.94)
Total: 36976 W: 7713 L: 7661 D: 21602
sprt @ 15+0.05 th 1 LMR improving: try best of both worlds
07-08-13 pe mc_prune_1 diff
LLR: -2.97 (-2.94,2.94)
Total: 14633 W: 2985 L: 3052 D: 8596
sprt @ 15+0.05 th 1 allow move count pruning at higher depth if eval is not improving
03-08-13 jk retireMinSplitDepthTwea diff
LLR: -2.97 (-2.94,2.94)
Total: 7484 W: 1188 L: 1309 D: 4987
sprt @ 60+0.05 th 3 Always use minimumSplitDepth of 7. Test with three threads. Regression test.
02-08-13 tv king_threats diff
LLR: -1.20 (-2.94,2.94)
Total: 42589 W: 9113 L: 9034 D: 24442
sprt @ 15+0.05 th 1 Another variation of Eelco's King Threat parameters
07-08-13 tv LMRHalvePly diff
LLR: -2.94 (-2.94,2.94)
Total: 4004 W: 775 L: 871 D: 2358
sprt @ 15+0.05 th 1 Test Gary's Counter_Cut + Marco 1/2 ply LMR against Marco 1/2 ply LMR
07-08-13 tv 281ce13 diff
LLR: -0.43 (-2.94,2.94)
Total: 229 W: 42 L: 57 D: 130
sprt @ 15+0.05 th 1 Test Gary's Counter_Cut + Marco 1/2 ply LMR against Marco 1/2 ply LMR
06-08-13 mc 86eca0561f8aac6647f9d3c diff
ELO: 2.22 +-2.9 (95%) LOS: 93.3%
Total: 20000 W: 3732 L: 3604 D: 12664
20000 @ 60+0.05 th 1 Pitthe2lmr improvingversions
02-08-13 ra CounterMoves_A diff
LLR: -2.97 (-2.94,2.94)
Total: 66872 W: 14095 L: 14012 D: 38765
sprt @ 15+0.05 th 1 Try to fix bug in null move counter moves
30-07-13 gl lmr_tweak diff
LLR: -2.95 (-2.94,2.94)
Total: 16437 W: 3434 L: 3495 D: 9508
sprt @ 15+0.05 th 1 Take 2: Reduce LMR less
29-07-13 gl danger_logistic diff
LLR: -2.96 (-2.94,2.94)
Total: 14085 W: 2990 L: 3058 D: 8037
sprt @ 15+0.05 th 1 Try a logistic function for KingDanger
29-07-13 gl counter_cut diff
LLR: 2.97 (-2.94,2.94)
Total: 51809 W: 10997 L: 10740 D: 30072
sprt @ 15+0.05 th 1 Reduce countermoves less in cut nodes
05-08-13 mc master diff
ELO: 50.95 +-3.2 (95%) LOS: 100.0%
Total: 20000 W: 5968 L: 3056 D: 10976
20000 @ 60+0.05 th 1 Regression test with Eelco's pruning when not improving
05-08-13 pe decrease_improving diff
LLR: -2.97 (-2.94,2.94)
Total: 36576 W: 7695 L: 7699 D: 21182
sprt @ 15+0.05 th 1 decrease pruning if eval improving
05-08-13 mc search diff
LLR: 2.97 (-2.94,2.94)
Total: 12992 W: 2491 L: 2318 D: 8183
sprt @ 60+0.05 th 1 Reschedule at long TC: Increase half ply lmr when not improving: take 1
05-08-13 tv PiecePawn diff
LLR: -2.97 (-2.94,2.94)
Total: 7680 W: 1609 L: 1696 D: 4375
sprt @ 15+0.05 th 1 Final test, no bonus on first rank
05-08-13 gl search diff
LLR: 2.95 (-2.94,2.94)
Total: 8748 W: 1689 L: 1539 D: 5520
sprt @ 60+0.05 th 1 (Marco Test) Long TC: Increase ONE_PLY lmr when not improving: take 2
04-08-13 mc search diff
LLR: 2.96 (-2.94,2.94)
Total: 37960 W: 8129 L: 7911 D: 21920
sprt @ 15+0.05 th 1 Increase half ply lmr when not improving: take 3
04-08-13 tv PiecePawn diff
LLR: 1.55 (-2.94,2.94)
Total: 48209 W: 9070 L: 8764 D: 30375
sprt @ 60+0.05 th 1 Rewrite/cleanup of pawn in front of minor as suggested by Marco