Stockfish Testing Queue

Finished - 56551 tests

19-09-15 Viz mgComplNonZeroing1 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 41975 W: 9210 L: 9168 D: 23597
sprt @ 10+0.1 th 1 Take 1
19-09-15 MJZ SF_formula1 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 4637 W: 945 L: 1086 D: 2606
sprt @ 10+0.1 th 1 SF new formula - 32 / 18 / 40 +2x (R+Q >1)
19-09-15 MJZ SF_formula2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 15659 W: 3378 L: 3465 D: 8816
sprt @ 10+0.1 th 1 SF new formula - 32 / 16 / 40 +2x (R+Q >1)
19-09-15 Viz pseudoAttackers2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 23602 W: 5130 L: 5178 D: 13294
sprt @ 10+0.1 th 1 Take 2
19-09-15 MJZ SF_formula2 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 8901 W: 1922 L: 2043 D: 4936
sprt @ 10+0.1 th 1 SF new formula - 32 / 12 / 40 +2x (R+Q >1)
19-09-15 Viz pseudoAttackers1 diff
LLR: -2.94 (-2.94,2.94) [0.50,4.50]
Total: 15422 W: 3276 L: 3364 D: 8782
sprt @ 10+0.1 th 1 Take 1
19-09-15 31m Viz_RonKQ diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 16858 W: 3667 L: 3748 D: 9443
sprt @ 10+0.1 th 1 With @xoto's tweak included
19-09-15 Viz spaceDblPawn1 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 14363 W: 3113 L: 3207 D: 8043
sprt @ 10+0.1 th 1 Take 1
19-09-15 xot mat5 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 38285 W: 8406 L: 8382 D: 21497
sprt @ 10+0.1 th 1 Lower scale factor if rooks still on, 43,4*.
19-09-15 31m Viz_RonKQ diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 57080 W: 12544 L: 12427 D: 32109
sprt @ 10+0.1 th 1 Respin this version after the simplification.
19-09-15 Viz oppositeComplexity1 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 32264 W: 7051 L: 7057 D: 18156
sprt @ 10+0.1 th 1 Take 1
19-09-15 xot mat5^ diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 37011 W: 8096 L: 8078 D: 20837
sprt @ 10+0.1 th 1 Lower scale factor if majors still on, 43,4*. Testing against Stephane's scale_factor8 patch
19-09-15 31m combo_09_15_19 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 20268 W: 4411 L: 4495 D: 11362
sprt @ 10+0.1 th 1 Drop the king danger tweak.
19-09-15 31m combo_09_15_19^ diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 20656 W: 4488 L: 4571 D: 11597
sprt @ 10+0.1 th 1 Drop the search tweak.
19-09-15 xot mat4 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 20873 W: 4594 L: 4655 D: 11624
sprt @ 10+0.1 th 1 Lower scale factor for rooks still on.
19-09-15 xot mat4 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 35461 W: 7864 L: 7853 D: 19744
sprt @ 10+0.1 th 1 Lower scale factor for majors still on.
19-09-15 sg tune_complex_eval diff
46413/100000 iterations
97292/200000 games played
200000 @ 20+0.2 th 1 So the first tuned values are really bad. I had done my first tuning with C=30 which seems to high. So try a second tuning with C=5 from zero values. Everything else keeps the same.
19-09-15 Viz mgInitNpm3 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 22040 W: 4792 L: 4848 D: 12400
sprt @ 10+0.1 th 1 Take 3
19-09-15 vdv averageCompletedDepth2 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 51976 W: 11476 L: 11384 D: 29116
sprt @ 10+0.1 th 1 Take b01.
19-09-15 31m combo_09_15_19 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 23644 W: 5106 L: 5177 D: 13361
sprt @ 10+0.1 th 1 Merge some recent tweaks by @xoto into an already-yellow combo.
19-09-15 vdv averageCompletedDepth2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 39126 W: 8615 L: 8586 D: 21925
sprt @ 10+0.1 th 1 Take b02.
19-09-15 Viz mgInitNpm1 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 42186 W: 9234 L: 9191 D: 23761
sprt @ 10+0.1 th 1 Take 1
19-09-15 vdv averageCompletedDepth2^ diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 36723 W: 7992 L: 7976 D: 20755
sprt @ 10+0.1 th 1 averageCompletedDepth and use it. Take 42.
19-09-15 Viz mgInitNpm2 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 22918 W: 4962 L: 5014 D: 12942
sprt @ 10+0.1 th 1 Take 2
19-09-15 vdv averageCompletedDepth2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 21525 W: 4693 L: 4751 D: 12081
sprt @ 10+0.1 th 1 Take 44.
19-09-15 vdv averageCompletedDepth2^ diff
LLR: -2.94 (-2.94,2.94) [0.50,4.50]
Total: 19716 W: 4229 L: 4296 D: 11191
sprt @ 10+0.1 th 1 Take 43.
19-09-15 Viz ss1statScoreCmhp1 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 20356 W: 4420 L: 4484 D: 11452
sprt @ 10+0.1 th 1 Take 1
19-09-15 Viz weakQKR1 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 15980 W: 3472 L: 3558 D: 8950
sprt @ 10+0.1 th 1 Take 1
19-09-15 MJZ SF_formula2 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 10679 W: 2338 L: 2450 D: 5891
sprt @ 10+0.1 th 1 SF new formula - 34 / 14 / 42 +2x (R+Q >1)
19-09-15 Viz PvNodeBestMoveCntLMR1 diff
LLR: -2.94 (-2.94,2.94) [0.50,4.50]
Total: 12750 W: 2753 L: 2854 D: 7143
sprt @ 10+0.1 th 1 Take 1
19-09-15 Viz weakKRblockers1 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 10487 W: 2259 L: 2372 D: 5856
sprt @ 10+0.1 th 1 Take 1
19-09-15 Viz weakQKR2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 8621 W: 1860 L: 1982 D: 4779
sprt @ 10+0.1 th 1 Take 2
19-09-15 sg tune_complex_eval diff
94166/100000 iterations
196828/200000 games played
200000 @ 20+0.2 th 1 Clamp values to a range and use less games. Hope its works. Experiment: try something new and "train" a simple CNN as additional eval term with SPSA with 200K games (probably several runs are needed but lets see). There are 143 parameters (I hope this is not much). I measured a speed loose of around 3-4% in comparison to master. Comments and discussion are welcome.
19-09-15 sg complex_eval diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 1085 W: 152 L: 308 D: 625
sprt @ 10+0.1 th 1 Test tuned nn values after 182K games to get a first impression of progress but i think we need several million games.
19-09-15 Viz bishopPairBlockedCenter diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 11825 W: 2537 L: 2643 D: 6645
sprt @ 10+0.1 th 1 Take 3
19-09-15 MJZ SF_formula2 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 26897 W: 5886 L: 5918 D: 15093
sprt @ 10+0.1 th 1 SF new formula - 32 / 14 / 42 +2x (R+Q >1)
19-09-15 Viz bishopPairBlockedCenter diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 6640 W: 1418 L: 1550 D: 3672
sprt @ 10+0.1 th 1 Take 4
19-09-15 Viz bishopPairBlockedCenter diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 21387 W: 4704 L: 4763 D: 11920
sprt @ 10+0.1 th 1 Take 1
19-09-15 MJZ SF_formula1 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 16485 W: 3534 L: 3617 D: 9334
sprt @ 10+0.1 th 1 SF new formula - 30 / 14 / 42 +2x (R+Q >1)
19-09-15 Viz bishopPairBlockedCenter diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 4486 W: 906 L: 1048 D: 2532
sprt @ 10+0.1 th 1 Take 2
19-09-15 xot initnpm4 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 15054 W: 3218 L: 3308 D: 8528
sprt @ 10+0.1 th 1 Initiative bonus if npm < 2050.
19-09-15 xot initnpm4 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 14408 W: 3140 L: 3233 D: 8035
sprt @ 10+0.1 th 1 Initiative npm bonus if npm < 1025.
19-09-15 MJZ SF_formula2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 7908 W: 1729 L: 1855 D: 4324
sprt @ 10+0.1 th 1 SF new formula - 28 / 14 / 42
19-09-15 MJZ SF_formula1 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 6390 W: 1371 L: 1504 D: 3515
sprt @ 10+0.1 th 1 SF new formula - 28 / 14 / 44
19-09-15 sni scale_factor8 diff
LLR: 2.95 (-2.94,2.94) [0.00,3.50]
Total: 30746 W: 5171 L: 4911 D: 20664
sprt @ 60+0.6 th 1 LTC: Take 3 of Scale down scale factor when shuffling
19-09-15 sni scale_factor8 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 54317 W: 11936 L: 11833 D: 30548
sprt @ 10+0.1 th 1 Take 4
19-09-15 Viz spaceWild3 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 5561 W: 1191 L: 1329 D: 3041
sprt @ 10+0.1 th 1 Take 3
19-09-15 sg scale_factor diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 51856 W: 11329 L: 11239 D: 29288
sprt @ 10+0.1 th 1 Reduce scale factor by one if material equal.
19-09-15 sni scale_factor8^ diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 48585 W: 10526 L: 10453 D: 27606
sprt @ 10+0.1 th 1 Scale down scale factor when shuffling
19-09-15 MJZ SF_formula1 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 8355 W: 1764 L: 1887 D: 4704
sprt @ 10+0.1 th 1 SF new formula - 28 / 14 / 40