Stockfish Testing Queue

Finished - 30366 tests

13-09-21 gli lmr diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 10729 W: 2172 L: 2036 D: 6521
sprt @ 15+0.05 th 1 Fix all/cut node when no reductions
13-09-21 gli lmrfix diff
LLR: -1.32 (-2.94,2.94) [-1.50,4.50]
Total: 38172 W: 7564 L: 7507 D: 23101
sprt @ 15+0.05 th 1 Fix all/cut definition when no LMR reduction
13-09-21 uri prune_more_with_less_ma diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 5797 W: 1095 L: 1185 D: 3517
sprt @ 15+0.05 th 1 I divided the wrong value in previous try. trying again to prune more based on value
13-09-21 rst static_threats diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 12535 W: 2432 L: 2504 D: 7599
sprt @ 15+0.05 th 1 extra bonus if we have many threats
13-09-21 mco search diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 13800 W: 2636 L: 2705 D: 8459
sprt @ 15+0.05 th 1 Retest Ralph idea of reducing quiets with no history with a much simpler implementation
13-09-21 mco search diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 11128 W: 2110 L: 2186 D: 6832
sprt @ 15+0.05 th 1 Reduce more quiet moves with non-positive history
13-09-21 pec lmr1 diff
LLR: -2.98 (-2.94,2.94) [-1.50,4.50]
Total: 11833 W: 2253 L: 2328 D: 7252
sprt @ 15+0.05 th 1 Decrease reduction for counter moves
13-09-22 gli lmr diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 49211 W: 8236 L: 8103 D: 32872
sprt @ 60+0.05 th 1 Long TC: Fix all/cut node when no reductions
13-09-22 mco search diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 20759 W: 4028 L: 4078 D: 12653
sprt @ 15+0.05 th 1 Reduce more quiets with negative score
13-09-22 mco search diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 3402 W: 708 L: 593 D: 2101
sprt @ 15+0.05 th 1 Reduce negative quiets by ONE_PLY / 2
13-09-22 mco time diff
ELO: 1.71 +-2.2 (95%) LOS: 93.8%
Total: 40000 W: 8325 L: 8128 D: 23547
40000 @ 10+0.05 th 1 Alexandre tweak to reduce time losses: tested at very short TC against original master
13-09-22 mco time diff
ELO: -0.29 +-2.2 (95%) LOS: 39.7%
Total: 40000 W: 8010 L: 8043 D: 23947
40000 @ 10+0.05 th 1 Alexandre tweak to reduce time losses: tested at very short TC against current master
13-09-22 mco search diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 8368 W: 1684 L: 1555 D: 5129
sprt @ 15+0.05 th 1 Reduce very negative quiets by ONE_PLY / 2
13-09-22 mco fix_lmr_cut_all diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 12301 W: 2372 L: 2445 D: 7484
sprt @ 15+0.05 th 1 Fix node type in LMR. From Gary's patch. Test it alone because it is a fix.
13-09-22 rst lmr_bad_quiets diff
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 5585 W: 1144 L: 1023 D: 3418
sprt @ 15+0.05 th 1 reduce bad quiets by 1/2 ply after fix
13-09-22 mco fix_lmr_cut_all diff
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 4333 W: 696 L: 778 D: 2859
sprt @ 60+0.05 th 1 LTC: Fix node type in LMR. Test it alone because it is a fix. If it fails also at long TC Gary's patch cannot be committed as is.
13-09-22 ren my_change diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 9275 W: 1784 L: 1865 D: 5626
sprt @ 15+0.05 th 1 Condition optimization. Increase the probability of break the condition ("!Signals.stop" will be false more often than "++depth <= MAX_PLY").
13-09-22 jos koutpost diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 23618 W: 4548 L: 4591 D: 14479
sprt @ 15+0.05 th 1 Another attempt to slightly increase bonus for Knight outposts near the enemy king.
13-09-22 rst search diff
LLR: 2.93 (-2.94,2.94) [0.00,6.00]
Total: 23379 W: 3972 L: 3759 D: 15648
sprt @ 60+0.05 th 1 LTC: Reduce negative quiets by ONE_PLY / 2
13-09-22 rst history_reductions diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 33038 W: 6411 L: 6428 D: 20199
sprt @ 15+0.05 th 1 Reduce negative quiets by ONE_PLY / 2, take 2
13-09-23 pec lmr1 diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 9591 W: 1989 L: 1855 D: 5747
sprt @ 15+0.05 th 1 reduce less on counter move
13-09-23 jos outpost_eg diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 20430 W: 3933 L: 3984 D: 12513
sprt @ 15+0.05 th 1 Halve outpost bonus in endgame score. Outposts are more valuable in midgame. Right?
13-09-23 mco reordering diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 4607 W: 852 L: 946 D: 2809
sprt @ 15+0.05 th 1 Reorder quiets after first bunch has been tried: history is changed in the meantime.
13-09-23 rst futility_pruning diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 14277 W: 2699 L: 2767 D: 8811
sprt @ 15+0.05 th 1 don't prune killers
13-09-23 rst futility_pruning diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 10955 W: 2078 L: 2155 D: 6722
sprt @ 15+0.05 th 1 do not prune countermoves
13-09-23 rst lmr1 diff
LLR: -2.97 (-2.94,2.94) [0.00,6.00]
Total: 32199 W: 5482 L: 5429 D: 21288
sprt @ 60+0.05 th 1 LTC: reduce less on counter move
13-09-23 gli lmr diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 8187 W: 2166 L: 2024 D: 3997
sprt @ 15+0.05 th 1 LMR bad captures one ply again, after the quiet move reduction patch
13-09-23 gli lmr_quiet diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 6553 W: 1423 L: 1514 D: 3616
sprt @ 15+0.05 th 1 Less LMR reductions, but more aggressive on LMR for quiet
13-09-23 gli lmr_quiet diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 53703 W: 11061 L: 10803 D: 31839
sprt @ 15+0.05 th 1 Take 2: Bump up PV LMR formula as well
13-09-23 gli master diff
ELO: 34.03 +-2.1 (95%) LOS: 100.0%
Total: 40000 W: 9354 L: 5449 D: 25197
40000 @ 60+0.05 th 1 Regression test against SF4 after TC changes and LMR quiet
13-09-23 gli lmr diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 4335 W: 732 L: 814 D: 2789
sprt @ 60+0.05 th 1 Long TC: LMR bad captures one ply again, after the quiet move reduction patch
13-09-23 pec lmr diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 13792 W: 2913 L: 2982 D: 7897
sprt @ 15+0.05 th 1 pv lmr formula change by itself
13-09-24 mco reordering diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 3483 W: 703 L: 802 D: 1978
sprt @ 15+0.05 th 1 Increase bad quiets ordering threshold
13-09-24 mco reordering diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 29567 W: 5714 L: 5740 D: 18113
sprt @ 15+0.05 th 1 Don't pollute history with low depth entries
13-09-24 Fis fuzzy_tt_cluster_test diff
ELO: -1.01 +-2.9 (95%) LOS: 25.2%
Total: 20000 W: 3726 L: 3784 D: 12490
20000 @ 15+0.05 th 1 testing overlapping clusters idea. starting with same size 4 as before.
13-09-24 rst counterfix diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 22022 W: 4261 L: 4308 D: 13453
sprt @ 15+0.05 th 1 validity check on countermoves update
13-09-24 rap bishop_pair diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 21096 W: 3985 L: 4035 D: 13076
sprt @ 15+0.05 th 1 Increase bishop pair bonus by 10%
13-09-24 mco reordering diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 3848 W: 720 L: 816 D: 2312
sprt @ 15+0.05 th 1 Remove bad quiets ordering threshold
13-09-24 rap history diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 21876 W: 4205 L: 4252 D: 13419
sprt @ 15+0.05 th 1 Extra reduce history update for lower than 4 depth
13-09-24 Fis fuzzy_tt_cluster_test diff
ELO: -0.05 +-3.0 (95%) LOS: 48.6%
Total: 20000 W: 3801 L: 3804 D: 12395
20000 @ 15+0.05 th 1 Testing/tuning overlapping clusters idea. This time with cluster size 3.
13-09-24 rap danger diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 24209 W: 4691 L: 4732 D: 14786
sprt @ 15+0.05 th 1 Don't prune negative SEE captures near the king
13-09-24 rap danger_B diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 18902 W: 3639 L: 3694 D: 11569
sprt @ 15+0.05 th 1 Don't prune captures near the king. Take 2: in qsearch
13-09-24 rst lmr_hist diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 3063 W: 562 L: 661 D: 1840
sprt @ 15+0.05 th 1 Adjust reductions according to reductions history
13-09-24 pec lmr_quiet diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 27696 W: 4550 L: 4521 D: 18625
sprt @ 60+0.05 th 1 LTC for Gary. Take 2: Bump up PV LMR formula as well
13-09-24 pec koutpost10 diff
LLR: -2.97 (-2.94,2.94) [0.00,6.00]
Total: 11616 W: 1950 L: 1997 D: 7669
sprt @ 60+0.05 th 1 LTC Increase Knight outpost bonus by 10% towards the short-castled king.
13-09-25 pec space diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 8595 W: 1573 L: 1656 D: 5366
sprt @ 15+0.05 th 1 Retune. Previous test seems to be contadicting
13-09-25 mco reordering^ diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 10967 W: 2053 L: 2130 D: 6784
sprt @ 15+0.05 th 1 Tune bad quiets ordering threshold: take 1
13-09-25 mco reordering diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 42551 W: 8279 L: 8270 D: 26002
sprt @ 15+0.05 th 1 Tune bad quiets ordering threshold: take 2
13-09-25 rap static_null_A diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 11677 W: 2329 L: 2404 D: 6944
sprt @ 15+0.05 th 1 Add margins in static null move pruning
13-09-25 jos kingRing diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 8333 W: 1712 L: 1797 D: 4824
sprt @ 15+0.05 th 1 Since queue is currently empty, give it a last try. This time extend KingRing one additional rank.