Stockfish Testing Queue

Finished - 2836 tests

11-12-17 vd fixStaticEval diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 104947 W: 18897 L: 18919 D: 67131
sprt @ 10+0.1 th 1 stc, take 2
12-12-17 Ro SSOC diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 31958 W: 4108 L: 4002 D: 23848
sprt @ 60+0.6 th 1 Yet another simplification
13-12-17 sg update_stats4 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 14007 W: 2638 L: 2453 D: 8916
sprt @ 10+0.1 th 1 Greater stats update if fail low at an expected pv or cut node for previous move
12-12-17 sg update_stats diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 60657 W: 11130 L: 10747 D: 38780
sprt @ 10+0.1 th 1 More stats bonus if best move found at expected ALL node.
12-12-17 pb enhanced_verify diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 41899 W: 5359 L: 5265 D: 31275
sprt @ 60+0.6 th 1 LTC: Enhance verification search for better detecting of zugzwang. [-3;1] bounds, see commit notes.
12-12-17 Ro SSOC diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 44211 W: 8009 L: 7928 D: 28274
sprt @ 10+0.1 th 1 Yet another simplification
12-12-17 fa pbk1 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 11580 W: 2169 L: 1995 D: 7416
sprt @ 10+0.1 th 1 First shot at Pawn Behind King extra penalty (With the help of Rocky)
12-12-17 pb enhanced_verify diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 18220 W: 3379 L: 3253 D: 11588
sprt @ 10+0.1 th 1 STC: Enhance verification search for better detecting of zugzwang. [-3;1] bounds, see commit notes.
11-12-17 SC context diff
LLR: 4.24 (-2.94,2.94) [-3.00,1.00]
Total: 22933 W: 4291 L: 4105 D: 14537
sprt @ 10+0.1 th 1 And +25%, take 4.
03-12-17 sg master diff
ELO: 39.88 +-1.5 (95%) LOS: 100.0%
Total: 40000 W: 6377 L: 1806 D: 31817
40000 @ 180+1.8 th 1 VLTC for comparison (see https://groups.google.com/forum/#!topic/fishcooking/A78xxpyU8dI). Regression test until "Use constexpr when makes sense" (low throughput)
11-12-17 Ro SOC3 diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 20292 W: 3796 L: 3570 D: 12926
sprt @ 10+0.1 th 1 2nd try to improve, against latest green LTC
10-12-17 II stempo diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 9323 W: 1749 L: 1585 D: 5989
sprt @ 10+0.1 th 1 Define Tempo as Score.
09-12-17 Ro SimpleOtherCheck2 diff
LLR: 2.94 (-2.94,2.94) [-3.00,1.00]
Total: 26812 W: 3575 L: 3463 D: 19774
sprt @ 60+0.6 th 1 LTC for Take 2: use a popcount
09-12-17 Ro SimpleOtherCheck diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 125304 W: 22761 L: 22819 D: 79724
sprt @ 10+0.1 th 1 simpler other checks, removing three conditional and using some kingDanger units.
09-12-17 II contem diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 25041 W: 4710 L: 4467 D: 15864
sprt @ 10+0.1 th 1 Take 2: Larger change
09-12-17 Ro SimpleOtherCheck2 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 11705 W: 2217 L: 2080 D: 7408
sprt @ 10+0.1 th 1 Take 2: use a popcount 1/2 throughput
08-12-17 vd noVerify diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 31550 W: 4020 L: 3913 D: 23617
sprt @ 60+0.6 th 1 ltc
07-12-17 vd noVerify diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 80980 W: 14697 L: 14679 D: 51604
sprt @ 10+0.1 th 1 stc
07-12-17 El pawnEndgames diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 45819 W: 5910 L: 5822 D: 34087
sprt @ 60+0.6 th 1 Try to simplify a bit more (LTC)
07-12-17 El pawnEndgames diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 40129 W: 7270 L: 7182 D: 25677
sprt @ 10+0.1 th 1 Try to simplify a bit more
06-12-17 Vo pvExactExp diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 75161 W: 9548 L: 9497 D: 56116
sprt @ 60+0.6 th 1 LTC: Suggested to use simplified bounds.
03-12-17 sg master diff
ELO: 43.31 +-1.7 (95%) LOS: 100.0%
Total: 40000 W: 7367 L: 2406 D: 30227
40000 @ 60+0.6 th 1 Regression test until "Use constexpr when makes sense"
04-12-17 Vo pvExactExp diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 110053 W: 20233 L: 19700 D: 70120
sprt @ 10+0.1 th 1 See how this goes...
05-12-17 An tuning_late_move_prunin diff
LLR: 3.50 (-2.94,2.94) [0.00,4.00]
Total: 25111 W: 4672 L: 4401 D: 16038
sprt @ 10+0.1 th 1 tune late/move count pruning values based off a custom tuner
04-12-17 pb queenBetween diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 63448 W: 8287 L: 7965 D: 47196
sprt @ 60+0.6 th 1 LTC: Don't consider queen as check-blocker
04-12-17 pb queenBetween diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 22636 W: 4212 L: 3990 D: 14434
sprt @ 10+0.1 th 1 Don't consider queen as check-blocker
03-12-17 vd pawnEndgames2 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 38780 W: 4946 L: 4848 D: 28986
sprt @ 60+0.6 th 1 ltc (variant @Stefano80)
03-12-17 Ro SimplifyBD diff
LLR: 2.94 (-2.94,2.94) [-3.00,1.00]
Total: 10879 W: 1981 L: 1844 D: 7054
sprt @ 10+0.1 th 1 2 almost non functional changes testing for non regression
03-12-17 vd pawnEndgames2 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 37445 W: 6866 L: 6773 D: 23806
sprt @ 10+0.1 th 1 stc (variant @Stefano80)
01-12-17 II time2 diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 97450 W: 17882 L: 17393 D: 62175
sprt @ 10+0.1 th 1 Time management - half tuned, half guess
01-12-17 go remove_qmi diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 49659 W: 9029 L: 8957 D: 31673
sprt @ 10+0.1 th 1 2nd try on qmi removal. This time compensate a little bit more than on first try. I'm hoping for a tiny ELO gain.
30-11-17 sg safe_attacks4 diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 56688 W: 10486 L: 10135 D: 36067
sprt @ 10+0.1 th 1 Ok increase midgame value of ThreatByAttackOnQueen even further. Test against my passed safe_attacks2 patch.
30-11-17 ce entry_threat diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 16505 W: 3133 L: 2936 D: 10436
sprt @ 10+0.1 th 1 Endgame bonus when king is two moves away from a weak pawn
29-11-17 sn phased_contempt3 diff
ELO: 154.11 +-3.4 (95%) LOS: 100.0%
Total: 20000 W: 9242 L: 910 D: 9848
20000 @ 60+0.6 th 1 LTC: phased_contempt3 with contempt=40 against SF 7.
28-11-17 sn master diff
ELO: 121.20 +-2.9 (95%) LOS: 100.0%
Total: 20000 W: 7263 L: 556 D: 12181
20000 @ 60+0.6 th 1 LTC reference run of master against SF 7 (low throughput).
28-11-17 sg weak_queen diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 59960 W: 11004 L: 10624 D: 38332
sprt @ 10+0.1 th 1 Half difference
28-11-17 vd pliesFromNull3 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 12807 W: 2449 L: 2268 D: 8090
sprt @ 10+0.1 th 1 stc, take 3
27-11-17 ce master diff
ELO: 119.11 +-2.3 (95%) LOS: 100.0%
Total: 41000 W: 16584 L: 3054 D: 21362
40000 @ 10+0.1 th 1 Test current master Contempt=40 against SF7
27-11-17 sn phased_contempt3 diff
ELO: 139.73 +-2.6 (95%) LOS: 100.0%
Total: 40000 W: 19277 L: 4005 D: 16718
40000 @ 10+0.1 th 1 Now try higher contempt values: contempt=60, use contempt/2 for endgame. Test against SF 7.
27-11-17 sn phased_contempt3 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 83360 W: 15089 L: 15075 D: 53196
sprt @ 10+0.1 th 1 Non-regresion test against master for phased_contempt3 with contempt=0. I have added a prefetch in evaluate() to limit as much as possible the speed penalty of loading Eval::Contempt in the hot path. SPRT[-3..1] at STC against master.
26-11-17 sn phased_contempt3 diff
ELO: 134.31 +-2.5 (95%) LOS: 100.0%
Total: 40000 W: 18224 L: 3488 D: 18288
40000 @ 10+0.1 th 1 Use full contempt for endgame contempt. Test against SF 7.
26-11-17 sn phased_contempt3 diff
ELO: 139.06 +-2.5 (95%) LOS: 100.0%
Total: 40000 W: 18518 L: 3312 D: 18170
40000 @ 10+0.1 th 1 Use contempt/2 for endgame contempt. Test against SF 7.
26-11-17 sn phased_contempt3 diff
ELO: 136.47 +-2.5 (95%) LOS: 100.0%
Total: 40000 W: 18488 L: 3537 D: 17975
40000 @ 10+0.1 th 1 Use contempt/4 for endgame contempt. Test against SF 7.
25-11-17 sg dynamic_contempt_fast5 diff
ELO: 130.84 +-2.4 (95%) LOS: 100.0%
Total: 40000 W: 17345 L: 2955 D: 19700
40000 @ 10+0.1 th 1 Test variant og the optimized version with contempt=20 against SF 7. Add contempt before evaluate_initiative so even sign of initiative can change
26-11-17 sg safe_attacks2 diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 7093 W: 987 L: 846 D: 5260
sprt @ 60+0.6 th 1 LTC: Now try double bonus
25-11-17 sg dynamic_contempt_fast4 diff
ELO: 131.71 +-2.4 (95%) LOS: 100.0%
Total: 40000 W: 17398 L: 2921 D: 19681
40000 @ 10+0.1 th 1 Sorry had the wrong book (so the elo gain was so good). Use contempt also for lazy eval like in my original patch. Perhaps this explains the missing 2 elo from optimized to original version. Includes vondele's patch with constraint contempt for negative eval (gained two elo). Test with contempt=20 against SF 7
26-11-17 sg safe_attacks2 diff
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 8629 W: 1599 L: 1438 D: 5592
sprt @ 10+0.1 th 1 Now try double bonus
26-11-17 sg safe_attacks diff
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 11895 W: 2193 L: 2018 D: 7684
sprt @ 10+0.1 th 1 Bonus for safe slider attack threats on opponent queen
26-11-17 sn phased_contempt2 diff
ELO: 132.28 +-2.6 (95%) LOS: 100.0%
Total: 40000 W: 18561 L: 4027 D: 17412
40000 @ 10+0.1 th 1 Delay initiative() simplification effect by using contempt to shift the attack/defense point (based on ceebo's "phased_contempt" patch). Tested against SF 7.
24-11-17 sg dynamic_contempt_fast3 diff
ELO: 132.36 +-2.4 (95%) LOS: 100.0%
Total: 40000 W: 17485 L: 2943 D: 19572
40000 @ 10+0.1 th 1 Retest the optimized vesion with contempt=20 against SF 7. Park with prio -10, see https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/B2psTv9FfoU