Stockfish Testing Queue

Finished - 4334 tests

19-07-14 sg nullmove_update diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 12390 W: 2667 L: 2770 D: 6953
sprt @ 10+0.1 th 1 If null move produces a cutoff (only second case with verification search) update histories for the previous move (combine quiet and non-quiet update).
19-07-14 sg nullmove_update diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 1970 W: 355 L: 509 D: 1106
sprt @ 10+0.1 th 1 If null move produces a cutoff (only first case without verification search) update cont histories for previous quiet move.
19-07-14 sg probcut_update diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 21715 W: 4793 L: 4850 D: 12072
sprt @ 10+0.1 th 1 Fixed (wrong bonus): Update capture stats after probcut fail high.
19-07-14 sg probcut_update diff
LLR: -0.77 (-2.94,2.94) [0.50,4.50]
Total: 6031 W: 1296 L: 1309 D: 3426
sprt @ 10+0.1 th 1 Update capture stats after probcut fail high.
19-07-14 sg lmr_cmh diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 10184 W: 2225 L: 2339 D: 5620
sprt @ 10+0.1 th 1 More reduction in second stat score rule if contHist[5] < -25000.
19-07-14 sg tweakStatScore diff
LLR: -2.94 (-2.94,2.94) [0.00,4.00]
Total: 26252 W: 5793 L: 5853 D: 14606
sprt @ 10+0.1 th 1 Tweak my yellow test and increase stat score even more.
19-07-12 sg sbcoeff1 diff
LLR: -2.97 (-2.94,2.94) [-3.00,3.00]
Total: 17454 W: 2826 L: 2929 D: 11699
sprt @ 60+0.6 th 1 LTC: Experiment: Try to select the better passed version test Viz'z sbcoeff1 patch against my lmr_update3_PR (see my comment to my PR https://github.com/official-stockfish/Stockfish/pull/2236). Low TP. Park with prio -1 until Viz's patch passes finally LTC.
19-07-12 sg lmr_update3 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 92357 W: 20487 L: 20286 D: 51584
sprt @ 10+0.1 th 1 Use depth instead of newDepth for LMR stat update. Test against my passed lmr_update2_PR patch.
19-07-12 sg lmr_update3 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 71213 W: 15890 L: 15771 D: 39552
sprt @ 10+0.1 th 1 Use 3/4 bonus for stat update after LMR-research. Test against my passed lmr_update2_PR patch.
19-07-12 sg tweakStatScore diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 44967 W: 9904 L: 9891 D: 25172
sprt @ 10+0.1 th 1 Try the other direction and increase stat score.
19-07-12 sg tweakStatScore diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 34492 W: 7557 L: 7585 D: 19350
sprt @ 10+0.1 th 1 Lower stat score (half effect).
19-07-12 sg lmr_stat_score diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 27264 W: 5963 L: 5993 D: 15308
sprt @ 10+0.1 th 1 Less reduction if check move in second stat score rule.
19-07-12 sg tweakStatScore diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 15238 W: 3269 L: 3373 D: 8596
sprt @ 10+0.1 th 1 Lower stat score to compensate for change of history update.
19-07-12 sg qsearch^ diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 12942 W: 2819 L: 2920 D: 7203
sprt @ 10+0.1 th 1 Don't prune move only for futility value pruning in quience search if piece is attacked by pawn.
19-07-12 sg tweakStatScore diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 10900 W: 2349 L: 2470 D: 6081
sprt @ 10+0.1 th 1 After change of stat update restore old stat score LMR divisor of 20000.
19-07-12 sg qsearch diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 8364 W: 1797 L: 1920 D: 4647
sprt @ 10+0.1 th 1 Don't prune move only for futility base pruning in quience search if piece is attacked by pawn.
19-07-12 sg qsearch diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 13426 W: 2888 L: 2986 D: 7552
sprt @ 10+0.1 th 1 Don't prune move only for SEE pruning in quience search if piece is attacked by pawn.
19-07-12 sg qsearch diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 12539 W: 2773 L: 2876 D: 6890
sprt @ 10+0.1 th 1 Don't prune move only for futility pruning in quience search if piece is attacked by pawn.
19-07-12 sg qsearch diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 10181 W: 2193 L: 2307 D: 5681
sprt @ 10+0.1 th 1 Don't prune move in quience search if piece is attacked by pawn.
19-07-12 sg lmr_update3 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 16245 W: 2832 L: 2704 D: 10709
sprt @ 60+0.6 th 1 LTC: Use now full bonus for stat update after LMR-research. Test against my passed lmr_update2_PR patch. Test as simplification.
19-07-12 sg lmr_update3 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 31630 W: 6892 L: 6932 D: 17806
sprt @ 10+0.1 th 1 Use only quarter bonus for stat update after LMR-research. Test against my passed lmr_update2_PR patch.
19-07-12 sg lmr_update3 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 23684 W: 5255 L: 5137 D: 13292
sprt @ 10+0.1 th 1 Use now full bonus for stat update after LMR-research. Test against my passed lmr_update2_PR patch. Test as simplification.
19-07-11 sg lmr_update2 diff
LLR: 2.96 (-2.94,2.94) [0.50,4.50]
Total: 39657 W: 8966 L: 8604 D: 22087
sprt @ 10+0.1 th 1 Update continuation history for quiet move which failed high at LMR dependent on the reaearch result but use only half bonus.
19-07-11 sg lmr_update2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 37370 W: 8379 L: 8357 D: 20634
sprt @ 10+0.1 th 1 Update continuation history for quiet move which failed high at LMR and also failed high at research but use only half bonus.
19-07-11 sg lmr_update2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 32345 W: 7139 L: 7143 D: 18063
sprt @ 10+0.1 th 1 Update continuation history for quiet move which failed high at LMR but research failed but use only half bonus.
19-07-11 sg lmr_update2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 26150 W: 5773 L: 5808 D: 14569
sprt @ 10+0.1 th 1 Update continuation history for quiet move which failed high at LMR but research failed.
19-07-11 sg lmr_update2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 8746 W: 1889 L: 2010 D: 4847
sprt @ 10+0.1 th 1 Use depth of LMR search for stat bonus.
19-07-11 sg lmr_update diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 12929 W: 2809 L: 2910 D: 7210
sprt @ 10+0.1 th 1 Fixed (used wrongly ss-1 instead of ss): Update cont history for fail high queit moves in LMR.
19-07-11 sg lmr_update diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 653 W: 67 L: 224 D: 362
sprt @ 10+0.1 th 1 Update cont history for all quiet moves in LMR.
19-07-11 sg lmr_update diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 632 W: 77 L: 240 D: 315
sprt @ 10+0.1 th 1 Update cont history for fail low quiet moves in LMR.
19-07-11 sg lmr_update diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 966 W: 145 L: 305 D: 516
sprt @ 10+0.1 th 1 Update continuation history for quiet moves which fail low in LMR search.
19-07-11 sg tweakGoodCaptures diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 27949 W: 4736 L: 4800 D: 18413
sprt @ 60+0.6 th 1 LTC: Now increase the SEE threshold for good captures by around 16%. Test against VoyagerOne's passed capScoreT4 patch.
19-07-10 sg piece_stats diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 38716 W: 8664 L: 8636 D: 21416
sprt @ 10+0.1 th 1 Reverse order of updating stats and applying difference. Test against xoto's passed redsscombo1pr patch.
19-07-10 sg piece_stats diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 25861 W: 5728 L: 5764 D: 14369
sprt @ 10+0.1 th 1 Double effect by adding difference / 8. Test against xoto's passed redsscombo1pr patch.
19-07-10 sg tweakGoodCaptures diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 107397 W: 24163 L: 23571 D: 59663
sprt @ 10+0.1 th 1 Now increase the SEE threshold for good captures by around 16%. Test against VoyagerOne's passed capScoreT4 patch.
19-07-10 sg piece_stats diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 52483 W: 11711 L: 11615 D: 29157
sprt @ 10+0.1 th 1 Try another idea and collect average stat score per piece and increase stat score by a fraction (1/16) of the difference to average. Test against xoto's passed redsscombo1pr patch.
19-07-10 sg tweakGoodCaptures diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 16283 W: 3556 L: 3656 D: 9071
sprt @ 10+0.1 th 1 Now increase the SEE threshold for good captures by around 55%. Test against VoyagerOne's passed capScoreT4 patch.
19-07-10 sg tweakGoodCaptures diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 51256 W: 11361 L: 11322 D: 28573
sprt @ 10+0.1 th 1 Now increase the SEE threshold for good captures by around 33%. Test against VoyagerOne's passed capScoreT4 patch.
19-07-10 sg piece_stats diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 20423 W: 4552 L: 4615 D: 11256
sprt @ 10+0.1 th 1 Less reduction if stat score < 0 and stat score > average stat + 10000. Test against xoto's passed redsscombo1pr patch.
19-07-10 sg piece_stats diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 15796 W: 3553 L: 3639 D: 8604
sprt @ 10+0.1 th 1 Less reduction if stat score > average stat + 40000. Test against xoto's passed redsscombo1pr patch.
19-07-10 sg piece_stats diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 9614 W: 2036 L: 2153 D: 5425
sprt @ 10+0.1 th 1 More reduction if stat score < average stat - 35000. Test against xoto's passed redsscombo1pr patch.
19-07-10 sg piece_stats diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 27736 W: 6133 L: 6160 D: 15443
sprt @ 10+0.1 th 1 More reduction if stat score is far below average piece stat score. Test against xoto's passed redsscombo1pr patch.
19-07-10 sg tweakGoodCaptures diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 17051 W: 3667 L: 3764 D: 9620
sprt @ 10+0.1 th 1 Decrease the SEE threshold for good captures by around 22%. Test against VoyagerOne's passed capScoreT4 patch.
19-07-10 sg tweakGoodCaptures diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 15378 W: 3348 L: 3452 D: 8578
sprt @ 10+0.1 th 1 Decrease the SEE threshold for good captures by around 42%. Test against VoyagerOne's passed capScoreT4 patch.
19-07-10 sg piece_stats diff
LLR: -2.94 (-2.94,2.94) [0.50,4.50]
Total: 5990 W: 1283 L: 1418 D: 3289
sprt @ 10+0.1 th 1 This fails fast so try smallest weight of 1. Also change rounding mode by adding N/2. Test against xoto's passed redsscombo1pr patch.
19-07-10 sg piece_stats diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 678 W: 84 L: 245 D: 349
sprt @ 10+0.1 th 1 Half weight of average stats.
19-07-10 sg piece_stats diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 315 W: 24 L: 199 D: 92
sprt @ 10+0.1 th 1 Half period of average stats.
19-07-10 sg piece_stats diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 304 W: 16 L: 179 D: 109
sprt @ 10+0.1 th 1 Collect stats dependent on piece if a quiet move was good/bad. Use the relative change (piece stat - overall stat) to adjust stat score in LMR.
19-07-10 sg quiet_order^ diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 28036 W: 6215 L: 6240 D: 15581
sprt @ 10+0.1 th 1 Scale in quiet ordering main history down with depth (break even depth=16, less slope).
19-07-10 sg quiet_order diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 20073 W: 4476 L: 4541 D: 11056
sprt @ 10+0.1 th 1 Scale in quiet ordering main history down with depth (break even depth=16, constrain factor to [0.5, 1.5]).