Stockfish Testing Queue

Pending - 0 tests 0.0 hrs

None

Active - 0 tests

Finished - 886 tests

18-05-25 pb0 castleFlex diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11473 W: 2285 L: 2360 D: 6828
sprt @ 10+0.1 th 1 Bonus if enemy Queen is still on board and we can castle, take 2
18-05-25 pb0 castleFlex diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 14209 W: 2859 L: 2922 D: 8428
sprt @ 10+0.1 th 1 Bonus if enemy Queen is still on board and we can castle immediately
18-05-25 pb0 pawn_check diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16254 W: 3265 L: 3318 D: 9671
sprt @ 10+0.1 th 1 Extra bonus (ThreatByPawnPush) when pawn push threats K,Q or R
18-05-24 pb0 pawn_check diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 44213 W: 8865 L: 8790 D: 26558
sprt @ 10+0.1 th 1 Extra bonus (ThreatByPawnPush) when pawn push gives check
18-05-24 pb0 pawn_disco diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17041 W: 3325 L: 3375 D: 10341
sprt @ 10+0.1 th 1 Take 2: Kingdanger when enemy pawn can give discovered check either by capturing or pushing forward.
18-05-24 pb0 pawn_disco diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 21828 W: 4315 L: 4343 D: 13170
sprt @ 10+0.1 th 1 Additional Kingdanger when enemy pawn can give discovered check either by capturing or pushing forward.
18-05-24 pb0 pawncheck diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 17820 W: 3507 L: 3554 D: 10759
sprt @ 10+0.1 th 1 pawn check by captures
18-05-24 pb0 nmp_exp3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8934 W: 1755 L: 1842 D: 5337
sprt @ 10+0.1 th 1 another nmp experiment
18-05-24 pb0 nmp_exp diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7066 W: 1340 L: 1436 D: 4290
sprt @ 10+0.1 th 1 nmp experiment
18-05-24 pb0 nmp_exp diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4165 W: 779 L: 888 D: 2498
sprt @ 10+0.1 th 1 nmp experiment, take 2
18-05-24 pb0 checkExtension diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 25099 W: 5090 L: 5102 D: 14907
sprt @ 10+0.1 th 1 Extend discovering checks when king cannot take the moved piece
18-05-24 pb0 disco_see diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 24085 W: 4826 L: 4843 D: 14416
sprt @ 10+0.1 th 1 Handle discovering check captures in SEE, take 4
18-05-22 pb0 disco_see diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 23428 W: 4664 L: 4685 D: 14079
sprt @ 10+0.1 th 1 Handle discovering check captures in SEE, take 3
18-05-22 pb0 disco_see diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10133 W: 1988 L: 2070 D: 6075
sprt @ 10+0.1 th 1 Handle discovering check captures in SEE
18-05-22 pb0 discoGoodCapture diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12591 W: 2505 L: 2575 D: 7511
sprt @ 10+0.1 th 1 Take 2
18-05-22 pb0 discoGoodCapture diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8930 W: 1718 L: 1805 D: 5407
sprt @ 10+0.1 th 1 Always handle discovering checking capture as good capture
18-05-18 pb0 fatalMoves diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 20150 W: 4087 L: 4122 D: 11941
sprt @ 10+0.1 th 1 Help opponent to find easy mates early, take 2
18-05-18 pb0 fatalMoves^ diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10021 W: 1970 L: 2052 D: 5999
sprt @ 10+0.1 th 1 Help opponent to find easy mates early
18-05-18 pb0 best_is_draw diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6966 W: 1343 L: 1439 D: 4184
sprt @ 10+0.1 th 1 Use draw-move as ttMove in siblings, take 4
18-05-17 pb0 best_is_draw diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 36614 W: 7301 L: 7261 D: 22052
sprt @ 10+0.1 th 1 Use draw-move as ttMove in siblings, take 3 (still experimental)
18-05-18 pb0 reachForDraw diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 3841 W: 698 L: 808 D: 2335
sprt @ 10+0.1 th 1 Are killers better searched before good captures when aiming to a draw?
18-05-17 pb0 best_is_draw diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8831 W: 1733 L: 1821 D: 5277
sprt @ 10+0.1 th 1 Use draw-move as ttMove in siblings, take 2 (experimental stuff)
18-05-17 pb0 best_is_draw diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 11469 W: 2242 L: 2318 D: 6909
sprt @ 10+0.1 th 1 Use draw-move as ttMove in siblings when alpha < 0 and we have no ttMove
18-05-17 pb0 closeQueenChecks diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 28908 W: 5824 L: 5819 D: 17265
sprt @ 10+0.1 th 1 Queenchecks: weight closeness to king, take 2
18-05-17 pb0 closeQueenChecks diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 668 W: 99 L: 229 D: 340
sprt @ 10+0.1 th 1 Queenchecks: weight closeness to king
18-05-17 pb0 kingMob2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 38397 W: 7823 L: 7773 D: 22801
sprt @ 10+0.1 th 1 Take 4, give checks different weights according to king mobility
18-05-16 pb0 kingMob2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 19274 W: 3918 L: 3957 D: 11399
sprt @ 10+0.1 th 1 King mobility in danger, take 3
18-05-16 pb0 kingMob2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 14599 W: 2913 L: 2974 D: 8712
sprt @ 10+0.1 th 1 King mobility in danger, take 2
18-05-16 pb0 kingMob2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 2404 W: 460 L: 579 D: 1365
sprt @ 10+0.1 th 1 King mobility in danger
18-05-15 pb0 capAspiration diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18353 W: 3517 L: 3562 D: 11274
sprt @ 5+0.05 th 4 SMP: shift alpha & beta to multiplies of 4 on large aspiration windows. Aim is to obtain more efficiency at early TT cut-offs. Local test: W:100 L: 80 D:271
18-05-15 pb0 smp_realign diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 17778 W: 3417 L: 3464 D: 10897
sprt @ 5+0.05 th 4 smp experiment take 3
18-05-14 pb0 smp_realign diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 39763 W: 7775 L: 7723 D: 24265
sprt @ 5+0.05 th 4 smp experiment take 2, act on higher depths only
18-05-14 pb0 smp_realign diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 906 W: 152 L: 281 D: 473
sprt @ 5+0.05 th 4 smp experiment: update helper's remarkably behind in score and depth (fixed logic)
18-05-14 pb0 smp_realign diff
LLR: 0.23 (-2.94,2.94) [0.00,5.00]
Total: 5052 W: 924 L: 893 D: 3235
sprt @ 5+0.05 th 8 smp experiment: update helper's remarkably behind in score and depth
18-05-14 pb0 ttSaveProbcut3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5224 W: 1023 L: 1128 D: 3073
sprt @ 10+0.1 th 1 Take 5 and probably last on this idea
18-05-14 pb0 ttSaveProbcut3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 24253 W: 4879 L: 4895 D: 14479
sprt @ 10+0.1 th 1 Take 4: On probcuts store to TT, so that it can produce early cut-offs
18-05-14 pb0 ttSaveProbcut3^ diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 24354 W: 4865 L: 4881 D: 14608
sprt @ 10+0.1 th 1 Take 3: On probcuts store to TT, so that it can produce early cut-offs
18-05-11 pb0 ttSaveProbcut3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 64079 W: 12966 L: 12798 D: 38315
sprt @ 10+0.1 th 1 On probcuts store into TT in way that it can even produce early cut-offs
18-05-11 pb0 ttSaveProbcut3 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 24933 W: 5074 L: 5087 D: 14772
sprt @ 10+0.1 th 1 Take 2, this one looks even stronger at local test.
18-05-11 pb0 firstMove diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 3523 W: 650 L: 762 D: 2111
sprt @ 10+0.1 th 1 Take 2 Movepicker: instead to pass empty ttMove, use 'capture killers' with positive SEE. Use 2 killers and apply logic on qsearch too.
18-05-10 pb0 allNodesTTmove diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4319 W: 842 L: 951 D: 2526
sprt @ 10+0.1 th 1 Store the first tried move in TT on all-nodes (no bestMove & no ttMove) This way next time when the search crosses this node, this potential good move can be tried and may cause a cut-off without move-generation.
18-05-10 pb0 ttSaveProbCut^ diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 77973 W: 15829 L: 15597 D: 46547
sprt @ 10+0.1 th 1 Save probCut-move in tt, experiment 2
18-05-10 pb0 firstMove diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 3049 W: 594 L: 710 D: 1745
sprt @ 10+0.1 th 1 Movepicker ttMove: instead to pass MOVE_NONE, use killers in that case. We potentially obtain a cut-off without need to generate any move.
18-05-10 pb0 smp_special diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 501 W: 59 L: 188 D: 254
sprt @ 5+0.05 th 8 SMP experiment: if the rootPos isn't a clear win and has several moves (captures or checks), then let a special thread skip searching the currently best move. This solves zugzwang pos 8/6pp/1K6/N5P1/3N4/8/npn1P3/k7 w - - 0 1
18-05-10 pb0 ttSaveProbCut diff
LLR: -3.02 (-2.94,2.94) [0.00,5.00]
Total: 6601 W: 1286 L: 1387 D: 3928
sprt @ 10+0.1 th 1 Save probCut-move in tt, experiment 3
18-05-09 pb0 ttSaveProbCut diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 57241 W: 8541 L: 8453 D: 40247
sprt @ 60+0.6 th 1 Save probCut-move in tt (see comments on commit)
18-05-09 pb0 ttSaveProbCut diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 68374 W: 14038 L: 13592 D: 40744
sprt @ 10+0.1 th 1 Save probCut-move in tt
18-05-09 pb0 ttSaveProbCut diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13988 W: 2803 L: 2867 D: 8318
sprt @ 10+0.1 th 1 Save probCut-move in tt, fixed logic (thanks to Stefan)
18-05-09 pb0 captStat2Score diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 25177 W: 5069 L: 5081 D: 15027
sprt @ 10+0.1 th 1 Let capture statistics influence nmp Do null-search after capture only if stats < 25000 (~ 92%) Sorry if I did'nt format properly, it's not easy to get this readable.
18-05-09 pb0 captStat2Score diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 32338 W: 6515 L: 6494 D: 19329
sprt @ 10+0.1 th 1 Let capture statistics update ss->statScore too, take 3