Stockfish Testing Queue

Finished - 30521 tests

15-11-03 jos easymove_tweak^ diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18270 W: 3559 L: 3604 D: 11107
sprt @ 10+0.1 th 1 Spend even less time on recaptures. Take 2.
15-11-03 jos easymove_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 34061 W: 6564 L: 6539 D: 20958
sprt @ 10+0.1 th 1 Spend less time on easy moves which are likely to be recaptures. And a little more on other moves. Take 1. (Inspired by Hazzl's tests)
15-11-02 bin deepest_move diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 12614 W: 2050 L: 2225 D: 8339
sprt @ 10+0.1 th 7 Simply check for best score (pointed out by lucabrivio)
15-11-03 lbr easy diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 16545 W: 3138 L: 3327 D: 10080
sprt @ 10+0.1 th 1 No easy move. Slow Mover=70.
15-10-26 aji lazy_log_formula_b diff
LLR: 2.96 (-2.94,2.94) [0.00,4.00]
Total: 8559 W: 1145 L: 977 D: 6437
sprt @ 60+0.4 th 24 Running LTC since STC did well at 24 threads and STC 3 threads doesn't seem to be a regression after 10K games.
15-11-03 tvi mob diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10657 W: 2032 L: 2111 D: 6514
sprt @ 10+0.1 th 1 Non linear mobility
15-11-02 lbr tune diff
22392/40000 iterations
45223/80000 games played
80000 @ 10+0.1 th 1 tune PV instability
15-11-03 sg pinned_attacks diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 9164 W: 1693 L: 1819 D: 5652
sprt @ 10+0.1 th 1 Try a more general approach. Use all pinned attacks (except for mobility). Because only code is rearranged test with SPRT[0,4].
15-11-03 Fis no_main_research diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2876 W: 402 L: 513 D: 1961
sprt @ 10+0.1 th 7 Spare the main thread from re-searching fail low/high whenever possible.
15-11-02 Haz careful1 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 42613 W: 8247 L: 8183 D: 26183
sprt @ 10+0.1 th 1 be more careful with easyMoves when the easyMove is a capture
15-11-02 Haz careful2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 37961 W: 7402 L: 7358 D: 23201
sprt @ 10+0.1 th 1 spend more time when easyMove is a capture (take2)
15-11-02 SC init_tuned diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 42809 W: 8336 L: 8341 D: 26132
sprt @ 10+0.1 th 1 Maybe is worth a further attempt: extrapolate from values which failed yellow at STC. Take 4.
15-11-01 Roc OnlyOpenFile diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17098 W: 3295 L: 3345 D: 10458
sprt @ 10+0.1 th 1 Increase the RookOnOpenFile when there is only one open file
15-11-03 pec fm_time diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 19159 W: 3969 L: 3752 D: 11438
sprt @ 15 th 1 Take 3: Stricter first root move
15-11-03 Roc RookvsMinors diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10055 W: 1937 L: 2019 D: 6099
sprt @ 10+0.1 th 1 Remove squares attacked by Minors on our first 4 ranks (in the Rook mobility Area)
15-11-02 Voy Explorer diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 1451 W: 236 L: 358 D: 857
sprt @ 10+0.1 th 1 Explore nodes...that don't have stats.
15-11-02 mco no_timer diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 32969 W: 6720 L: 6620 D: 19629
sprt @ 2+0.05 th 7 Retire timer. Novel approach: first thread to reach limit wins (takes in account qsearch and no global counter, no now() call). Use atomics for shared variables
15-11-02 Voy StatFormula3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18054 W: 3458 L: 3504 D: 11092
sprt @ 10+0.1 th 1 Take 3
15-11-01 luc lazy_bestmoves diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 7925 W: 1207 L: 1335 D: 5383
sprt @ 10+0.1 th 7 Take into account depth and score in one run when picking the best thread; not tuned (just in case the queue dries up!)
15-11-02 Voy StatsTweak diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 28972 W: 5611 L: 5666 D: 17695
sprt @ 10+0.1 th 1 Stat Tweaks... (Fixed bench)
15-11-02 SC variance3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5464 W: 972 L: 1074 D: 3418
sprt @ 10+0.1 th 1 Variance stuff, take 3. Scale aspiration window with variance of root position.
15-11-01 sg history diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 95711 W: 13864 L: 14152 D: 67695
sprt @ 60+0.4 th 1 LTC: Simplification: Remove update value bound and use instead a capped value. Bench is the same as this kicks in for depth >= 18. Test against passed NewStatFormula patch.
15-11-02 bin deepest_move diff
ELO: -15.65 +-15.8 (95%) LOS: 2.6%
Total: 511 W: 59 L: 82 D: 370
5000 @ 10+0.1 th 17 (fixed) Quick test: Choose deepest thread with best score
15-11-02 lbr pv_instability diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6696 W: 1011 L: 1107 D: 4578
sprt @ 10+0.1 th 7 fixed version
15-11-02 bin deepest_move diff
ELO: -13.85 +-24.0 (95%) LOS: 12.9%
Total: 276 W: 42 L: 53 D: 181
5000 @ 5+0.05 th 17 Quick test: Choose best score between same depth threads (rather than the first deepest thread)
15-11-02 bin deepest_move2 diff
ELO: -2.60 +-11.2 (95%) LOS: 32.5%
Total: 1205 W: 192 L: 201 D: 812
10000 @ 5+0.05 th 17 Quick test on mstembera's idea of bestThread sorting, now with more threads to see if it actually works.
15-11-02 pec fm_time diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 31200 W: 6024 L: 6011 D: 19165
sprt @ 10+0.1 th 1 Take 2: Stricter first root move
15-11-02 sg king_safety diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 15906 W: 3028 L: 3084 D: 9794
sprt @ 10+0.1 th 1 Ignore king attacks which blocked by own pawns . Take 2
15-11-02 lbr pv_instability diff
LLR: -2.59 (-2.94,2.94) [0.00,5.00]
Total: 243 W: 10 L: 117 D: 116
sprt @ 10+0.1 th 7 thread neutral PV instability
15-11-02 sg king_safety diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8891 W: 1671 L: 1758 D: 5462
sprt @ 10+0.1 th 1 Ignore king attackers which only attacks own blocked pawns. Take 3
15-11-02 Voy StatFormula2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 17095 W: 3252 L: 3303 D: 10540
sprt @ 10+0.1 th 1 Take 2.
15-11-02 pec fm_time diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 28410 W: 5472 L: 5472 D: 17466
sprt @ 10+0.1 th 1 Enforce stricter first root move time
15-11-01 SC init_tuned diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 49900 W: 9650 L: 9630 D: 30620
sprt @ 10+0.1 th 1 Try to extract some useful information from the initiative tuning. Take 3 and then call it a day.
15-11-01 Voy StatFormula diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 25133 W: 4865 L: 4879 D: 15389
sprt @ 10+0.1 th 1 Try an interesting way to score stats at high depths.
15-11-01 jos flex_time diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 20966 W: 4057 L: 4090 D: 12819
sprt @ 10+0.1 th 1 Plan some more moves ahead in the opening phase, and less in the endgame.
15-11-01 sg king_safety diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 19963 W: 3868 L: 3905 D: 12190
sprt @ 10+0.1 th 1 Ignore king attackers which only attacks own pawns (so the attack is blocked)
15-11-01 lbr available diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 19709 W: 3705 L: 3899 D: 12105
sprt @ 10+0.1 th 1 take 2: try to remove the fudge factor 0.76 while at it
15-11-01 SC variance2 diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 9759 W: 1858 L: 1941 D: 5960
sprt @ 10+0.1 th 1 Take2 : add variance to futility margin.
15-11-01 Voy HistoryBonus diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 26885 W: 5247 L: 5253 D: 16385
sprt @ 10+0.1 th 1 Try bonus for History: d*d. While bonus for cmh remains: d*d+d-1
15-11-01 SC variance diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8233 W: 1580 L: 1670 D: 4983
sprt @ 10+0.1 th 1 An exotic (probably wrong) idea: interpret one term in evaluate initiative as position variance and avoid tuning at shallow depth if this variance is very high.
15-11-01 Fis deepest_move2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 27952 W: 4445 L: 4459 D: 19048
sprt @ 10+0.1 th 7 See commit notes.
15-11-01 Roc RookvsMinors diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 31474 W: 6089 L: 6075 D: 19310
sprt @ 10+0.1 th 1 the opposite: remove the non-empty squares controlled by opponent minor from the rook mobilityarea
15-10-31 sg history diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 164572 W: 31686 L: 31822 D: 101064
sprt @ 10+0.1 th 1 Simplification: Remove update value bound and use instead a capped value. Bench is the same as this kicks in for depth >= 18. Test against passed NewStatFormula patch.
15-11-01 mco no_timer diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 8149 W: 1367 L: 1227 D: 5555
sprt @ 10+0.1 th 7 Retire timer. Novel approach: first thread to reach limit wins (takes in account qsearch and no global counter, no now() call). Standard STC
15-10-31 lbr available diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 78128 W: 14881 L: 14861 D: 48386
sprt @ 10+0.1 th 1 Prevent Time.available() from returning more than the maximum time.
15-11-01 mco no_timer diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 15587 W: 3036 L: 2905 D: 9646
sprt @ 10+0.1 th 1 Retire timer. Novel approach: first thread to reach limit wins (takes in account qsearch and no global counter, no now() call). Standard STC single thread
15-11-01 mco deepest_move diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 192 W: 8 L: 151 D: 33
sprt @ 10+0.1 th 7 Dynamic deepest move: update and use best move after every iteration, not only before to return. This naturally extends mbootsector idea.
15-11-01 SC generalize_init2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2979 W: 515 L: 629 D: 1835
sprt @ 10+0.1 th 1 Game phase dependent initiative. Take 2.
15-11-01 Fis material_time diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7881 W: 1149 L: 1241 D: 5491
sprt @ 60+0.4 th 1 Adjust time usage based on remaining material.
15-11-01 lbr available diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 30550 W: 5929 L: 6145 D: 18476
sprt @ 15+0 th 1 Prevent Time.available() from returning more than the maximum time. Zero increment.