Stockfish Testing Queue

Finished - 40825 tests

15-09-04 Fis KMTune2 diff
34215/35000 iterations
69135/70000 games played
70000 @ 30+0.05 th 1 Tuning for secondary knight mobility. Take2
15-09-05 sni material diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 28121 W: 4400 L: 4174 D: 19547
sprt @ 60+0.05 th 1 LTC: Keep more pawns when attacking. Take 2 (half value).
15-09-05 sg flanking_storm2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 21142 W: 3924 L: 3958 D: 13260
sprt @ 15+0.05 th 1 Double flanking storm bonus
15-09-03 SC capture_gp diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 54203 W: 10257 L: 10145 D: 33801
sprt @ 15+0.05 th 1 Ok, optimal rank penalty does not depend on aggressor piece. Maybe depends on game phase? Take 1: opponent pawn count.
15-09-04 hxi symhist diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 46227 W: 7198 L: 7145 D: 31884
sprt @ 60+0.05 th 1 LTC: my attempt on history updates
15-09-04 sg flanking_storm2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 30969 W: 5770 L: 5761 D: 19438
sprt @ 15+0.05 th 1 Flanking storm bonus -50%
15-09-02 Fis simpleSave diff
LLR: -1.30 (-2.94,2.94) [-3.00,1.00]
Total: 38025 W: 7048 L: 7184 D: 23793
sprt @ 15+0.05 th 1 Try to remove a condition that is likely useless. 2MB STC Low throughput.
15-09-05 Dis pawn_front_minor diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12669 W: 2376 L: 2447 D: 7846
sprt @ 15+0.05 th 1 Take 2 : Remove bonus when minor piece behind enemy pawn.
15-09-04 sni material diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 26157 W: 4928 L: 4689 D: 16540
sprt @ 15+0.05 th 1 Keep more pawns when attacking. Take 2 (half value).
15-09-03 sni material diff
LLR: 0.07 (-2.94,2.94) [0.00,5.00]
Total: 36018 W: 6610 L: 6453 D: 22955
sprt @ 15+0.05 th 1 Keep more pawns when attacking
15-09-04 Voy AdvanceBonus diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 18402 W: 3341 L: 3388 D: 11673
sprt @ 15+0.05 th 1 Relative Rank bonus/penalty for history stats.
15-08-31 aji minsplitdepth2_max5 diff
LLR: -4.14 (-2.94,2.94) [0.00,5.00]
Total: 33001 W: 5007 L: 5054 D: 22940
sprt @ 15+0.05 th 7 minsplitdepth2 patch capped to a max of 5 slaves : STC Pls approve only after maintainers confirm this is the right configuration to retest the patch.
15-09-03 Dis pawn_front_minor diff
LLR: -2.08 (-2.94,2.94) [0.00,5.00]
Total: 29224 W: 5484 L: 5445 D: 18295
sprt @ 15+0.05 th 1 Give half the value that is given normally for a minor piece behind an enemy pawn. (+Fix thanks to joergoster)
15-09-04 jhe linear_stats1 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5627 W: 781 L: 881 D: 3965
sprt @ 60+0.05 th 1 Linear Stats LTC.
15-09-04 Roc BishopCoordination diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 6285 W: 1135 L: 1234 D: 3916
sprt @ 15+0.05 th 1 Bonus for bishop on adjacent diagonals. From a Mindbreaker idea.
15-09-03 Voy FLStat diff
LLR: 3.11 (-2.94,2.94) [0.00,5.00]
Total: 27498 W: 5267 L: 5014 D: 17217
sprt @ 15+0.05 th 1 Take 3: Try a smaller bonus. (Fixed Bench)
15-09-03 jhe linear_stats1 diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 22810 W: 4375 L: 4149 D: 14286
sprt @ 15+0.05 th 1 Try a linear stats bonus...
15-09-03 SC capture_gp diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 11617 W: 2135 L: 2210 D: 7272
sprt @ 15+0.05 th 1 Ok, optimal rank penalty does not depend on aggressor piece. Maybe depends on game phase? Take 3: opponent non pawn material.
15-09-04 Voy Killers diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5179 W: 947 L: 1051 D: 3181
sprt @ 15+0.05 th 1 Don't clear Killers.
15-09-03 SC capture_gp diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 4877 W: 847 L: 952 D: 3078
sprt @ 15+0.05 th 1 Ok, optimal rank penalty does not depend on aggressor piece. Maybe depends on game phase? Take 1: opponent piece count.
15-09-03 Mys mv diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 29585 W: 5581 L: 5635 D: 18369
sprt @ 15+0.05 th 1 Minors/pawns mg value increase
15-08-30 hxi symhist diff
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 114720 W: 21720 L: 21092 D: 71908
sprt @ 15+0.05 th 1 my attempt on history updates
15-09-03 Dis all_pawns_bishop_pen diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 24437 W: 4618 L: 4637 D: 15182
sprt @ 15+0.05 th 1 Changing bishop same pawn color penality to include all pawns.
15-09-02 Voy FLStat diff
LLR: -2.98 (-2.94,2.94) [0.00,5.00]
Total: 16025 W: 2470 L: 2532 D: 11023
sprt @ 60+0.05 th 1 Boss Level: Take 2
15-09-03 lbr cut diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 13198 W: 2428 L: 2497 D: 8273
sprt @ 15+0.05 th 1 all->cut: more bonus
15-09-03 Dis pawn_front_minor diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8121 W: 1541 L: 1632 D: 4948
sprt @ 15+0.05 th 1 Give half the value that is given normally for a minor piece behind an enemy pawn.
15-09-03 Voy IterationBonus diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 6062 W: 1105 L: 1205 D: 3752
sprt @ 15+0.05 th 1 Add an iteration bonus to stats.
15-09-01 Voy KingsCorner diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 48301 W: 9066 L: 8981 D: 30254
sprt @ 15+0.05 th 1 Based on data mining of piece-to_sq cutoff stats, the king to h1 or h8 shows an abnormally high count . I believe this over inflated entry, is messing with our stats, which is why he needed a low Max for stats. Note the bench change...local testing seems good.
15-09-02 Voy FLStat diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 21897 W: 4206 L: 3984 D: 13707
sprt @ 15+0.05 th 1 Take 2
15-09-02 Roc RookMobSimple diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10105 W: 1897 L: 1979 D: 6229
sprt @ 15+0.05 th 1 Take 2
15-09-02 jos combo1 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 12783 W: 2309 L: 2422 D: 8052
sprt @ 15+0.05 th 1 Combo of tuned weights and assorted eval bonuses.
15-08-29 Voy histNoRest diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 94960 W: 17901 L: 17726 D: 59333
sprt @ 15+0.05 th 1 Don't reset the entry...
15-09-02 Fis clusterAging diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 21217 W: 3922 L: 3956 D: 13339
sprt @ 15+0.05 th 1 Take 2 less aggressive version. 2MB STC
15-09-01 jos tuned_weights diff
ELO: 0.49 +-2.5 (95%) LOS: 64.9%
Total: 30000 W: 6043 L: 6001 D: 17956
30000 @ 9+0.03 th 1 Check some early values.
15-09-01 lbr history diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 4889 W: 837 L: 942 D: 3110
sprt @ 15+0.05 th 1 last try
15-08-30 Voy FLStat diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 72981 W: 13704 L: 13512 D: 45765
sprt @ 15+0.05 th 1 Give prior move a bonus if it caused a fail low.
15-09-01 Roc RookMob diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10238 W: 1855 L: 1937 D: 6446
sprt @ 15+0.05 th 1 Take 1
15-09-01 tvi pawns diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 7957 W: 1455 L: 1547 D: 4955
sprt @ 15+0.05 th 1 extend pawn endings
15-09-01 Voy WeibullCDF diff
LLR: -1.98 (-2.94,2.94) [0.00,5.00]
Total: 1140 W: 178 L: 258 D: 704
sprt @ 15+0.05 th 1 Take 2.
15-09-01 Fis clusterAging diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12119 W: 2228 L: 2301 D: 7590
sprt @ 15+0.05 th 1 Keep track of the most recently updated entry in a cluster and use that information in the replacement policy. 2MB STC Correct bench is 8349057
15-09-01 lbr history^ diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5208 W: 917 L: 1020 D: 3271
sprt @ 15+0.05 th 1 simple history update: max = 256
15-09-01 lbr history diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 3153 W: 546 L: 659 D: 1948
sprt @ 15+0.05 th 1 simple history update: max = 512
15-08-31 SC capture_tuned diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 26284 W: 4862 L: 4874 D: 16548
sprt @ 15+0.05 th 1 Almost final values from SPSA tuning for piece type dependent rank penalty in capture ordering.
15-09-01 Voy WeibullCDF diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9470 W: 1769 L: 1854 D: 5847
sprt @ 15+0.05 th 1 Use Weibull CDF to derive stats bonuses. This will have a smooth transition at all depths.
15-08-26 hxi newdepth diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 128352 W: 24388 L: 23950 D: 80014
sprt @ 15+0.05 th 1 use newDepth in step 13 and 15
15-08-30 Voy FailLowBonusAdj diff
LLR: -3.16 (-2.94,2.94) [0.00,5.00]
Total: 47224 W: 8891 L: 8819 D: 29514
sprt @ 15+0.05 th 1 Decrease bonus if search fails low.
15-08-31 SC return_if_4 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 1543 W: 198 L: 315 D: 1030
sprt @ 15+0.05 th 1 Why are we searching positions with 4 or less pieces on the board? If I understand it correctly, (almost all) those positions are correctly treated by evaluate(pos).
15-08-31 lbr tune diff
ELO: -8.01 +-3.1 (95%) LOS: 0.0%
Total: 21033 W: 4323 L: 4808 D: 11902
30000 @ 9+0.03 th 1 quick measure: simplify storm, tuned locally.
15-08-30 lbr tune diff
ELO: -7.98 +-2.6 (95%) LOS: 0.0%
Total: 30000 W: 6194 L: 6883 D: 16923
30000 @ 9+0.03 th 1 quick measure: simplify storm, untuned.
15-08-30 Roc PawnThreat3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18087 W: 3392 L: 3439 D: 11256
sprt @ 15+0.05 th 1 Let's try a PawnAttackThreat in the missing case where a pawn can safely push and attack thanks to Rook or Queen support from behind.