Stockfish Testing Queue

Finished - 4464 tests

19-06-15 MJZ PE-shuffle2 diff
LLR: -1.12 (-2.94,2.94) [0.00,3.50]
Total: 10187 W: 1729 L: 1767 D: 6691
sprt @ 60+0.6 th 1 Return draw if PV is exploring the same shuffling positions. filter d > 8. Spec LTC
19-06-14 MJZ PE-shuffle diff
ELO: 3.06 +-5.7 (95%) LOS: 85.5%
Total: 5000 W: 889 L: 845 D: 3266
5000 @ 60+0.6 th 1 Return draw if PV is exploring the same shuffling positions. 5k games to check the idea @LTC
19-06-14 31m tweak_ThreatOnPawn diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 21107 W: 3548 L: 3634 D: 13925
sprt @ 60+0.6 th 1 Merge new master and LTC: add mg and eg components.
19-06-14 xot pawnhash2 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 69737 W: 11906 L: 11884 D: 45947
sprt @ 60+0.6 th 1 LTC. Scale pawn hash with size of main TT. Now 2x master at STC (so 16x master at LTC, etc). Updated with suggestions from sg.
19-06-14 Elb closed_ini diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 33670 W: 5787 L: 5867 D: 22016
sprt @ 60+0.6 th 1 Increase initiative for closed positions, symmetric variant (LTC)(thanks locutus2)
19-06-13 MJZ NS-Reduc3 diff
LLR: -1.22 (-2.94,2.94) [0.00,3.50]
Total: 46953 W: 8055 L: 7994 D: 30904
sprt @ 60+0.6 th 1 Increase reduction at root if searched nodes < SN0 / 32 && depth > 16 - needs high depth search
19-06-14 Viz rankThreatsP1 diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 21688 W: 3646 L: 3761 D: 14281
sprt @ 60+0.6 th 1 LTC
19-06-14 vdv shuffleExtFix diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 52042 W: 8907 L: 8837 D: 34298
sprt @ 60+0.6 th 1 Fix progress issue.
19-06-13 MJZ Root-noSkip diff
LLR: -0.53 (-2.94,2.94) [-3.00,1.00]
Total: 29219 W: 4237 L: 4306 D: 20676
sprt @ 60+0.6 th 4 It seems that the cost of skipping LMR at root is too high at higher depths. This is a verification if it is true @LTC x4 threads (I will stop it if no improvement after 10k games)
19-06-13 MJZ PvNode-progrSkip2 diff
LLR: -0.81 (-2.94,2.94) [0.00,3.50]
Total: 13162 W: 2232 L: 2243 D: 8687
sprt @ 60+0.6 th 1 Progressively decrease the limit of skipping LMR at PvNode. 4 - depth / 5 - spec LTC
19-06-13 Lol king-danger-ring diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 40992 W: 6849 L: 6910 D: 27233
sprt @ 60+0.6 th 1 Take 3 LTC
19-06-13 xor rPp diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 99536 W: 17039 L: 16932 D: 65565
sprt @ 60+0.6 th 1 LTC: take 1.
19-06-12 MJZ Root-Candidates2 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 19789 W: 3293 L: 3413 D: 13083
sprt @ 60+0.6 th 1 Limit reduction to 1 ply for root "best" moves. Criteria searched nodes > 4 + first move / 10 - keep skipping 4 first moves - LTC (please accept if STC passes or let it Spec LTC)
19-06-12 MJZ Root-SkipLMR2 diff
LLR: -0.69 (-2.94,2.94) [0.00,3.50]
Total: 10333 W: 1756 L: 1768 D: 6809
sprt @ 60+0.6 th 1 Skip LMR for root "best" moves instead of 4 first ones. Criteria searched nodes > 4 + first move / 10, Spec LTC (I will stop it if no improvement after 10k games)
19-06-07 sg pawnsTable diff
LLR: 0.31 (-2.94,2.94) [0.00,4.00]
Total: 156117 W: 27046 L: 26519 D: 102552
sprt @ 60+0.6 th 1 Doubling size seems good for LTC but not good enough. So try a speculative LTC with even 4 times more pawn cache (this adds 5 MB memory usage per thread) (Low TP)
19-06-12 sg pawnsTable diff
LLR: 2.94 (-2.94,2.94) [0.00,4.00]
Total: 26340 W: 4524 L: 4286 D: 17530
sprt @ 60+0.6 th 1 The test with doubling sizeof pawn cache struggles (have set prio back to -1). So try a speculative LTC with even 8 times more pawn cache (this adds 12.25 MB memory usage per thread) (Low TP)
19-06-12 sg bishop_psqt diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 15096 W: 2559 L: 2664 D: 9873
sprt @ 60+0.6 th 1 LTC: Half compensation for bishop psqt.
19-06-09 vdv master diff
LLR: -1.09 (-2.94,2.94) [0.00,3.50]
Total: 5226 W: 687 L: 737 D: 3802
sprt @ 60+0.6 th 8 large hash test of ttPV reduction (see fishcooking discussion).
19-06-11 MJZ NS-timeReduc-simple2 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 21309 W: 3604 L: 3794 D: 13911
sprt @ 60+0.6 th 1 Time management simplification - LTC of http://tests.stockfishchess.org/tests/view/5cfec2510ebc5925cf09884b
19-06-11 pra ad1 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 48252 W: 8291 L: 8329 D: 31632
sprt @ 60+0.6 th 1 testing LTC for 79k yellow stc low throughput
19-06-11 Voy dcS diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 28428 W: 4873 L: 4765 D: 18790
sprt @ 60+0.6 th 1 LTC Simplify recent patch
19-06-11 sg knight_psqt diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 28479 W: 4818 L: 4880 D: 18781
sprt @ 60+0.6 th 1 LTC: Half compensation for knight psqt.
19-06-11 MJZ NS-timeReduc-simple3 diff
ELO: -4.10 +-5.6 (95%) LOS: 7.6%
Total: 5000 W: 822 L: 881 D: 3297
5000 @ 60+0.6 th 1 TM simplification - test 5k games with pow3, 0.36 - 1.8 @LTC
19-06-11 MJZ NS-timeReduc-simple3 diff
ELO: -6.80 +-7.1 (95%) LOS: 3.0%
Total: 3116 W: 494 L: 555 D: 2067
5000 @ 60+0.6 th 1 TM simplification - test 5k games with pow3, 0.38 - 1.9 @LTC
19-06-11 MJZ NS-timeReduc-simple3 diff
ELO: -6.44 +-6.4 (95%) LOS: 2.4%
Total: 3777 W: 598 L: 668 D: 2511
5000 @ 60+0.6 th 1 TM simplification - test 5k games with pow3, 0.32 - 1.9 @LTC
19-06-11 MJZ NS-timeReduc-simple diff
ELO: -3.92 +-6.0 (95%) LOS: 10.2%
Total: 4429 W: 751 L: 801 D: 2877
5000 @ 60+0.6 th 1 TM simplification - test 5k games with pow3, 0.36 - 1.9 @LTC
19-06-11 MJZ NS-timeReduc-simple diff
ELO: -7.09 +-6.9 (95%) LOS: 2.2%
Total: 3382 W: 557 L: 626 D: 2199
5000 @ 60+0.6 th 1 TM simplification - test 5k games with linear 0.36 - 1.1 @LTC
19-06-10 Fis PawnHash diff
LLR: -2.96 (-2.94,2.94) [0.50,3.50]
Total: 33043 W: 5565 L: 5664 D: 21814
sprt @ 60+0.6 th 1 Spec LTC to increase pawn cache pressure. 20% throughput. Try storing 2 entries per pawn hash table slot with a replacement policy.
19-06-10 MJZ NS-timeReduc-simple diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 17002 W: 2840 L: 3023 D: 11139
sprt @ 60+0.6 th 1 Simplify time management - LTC of http://tests.stockfishchess.org/tests/view/5cfeb0490ebc5925cf0984ec
19-06-10 Voy gcPruning3 diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 15469 W: 2555 L: 2688 D: 10226
sprt @ 60+0.6 th 1 ltc
19-06-10 Voy gcPruning3 diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 16116 W: 2641 L: 2772 D: 10703
sprt @ 60+0.6 th 1 LTC
19-06-04 fau FauziB20 diff
ELO: 0.71 +-2.6 (95%) LOS: 70.1%
Total: 20000 W: 3030 L: 2989 D: 13981
20000 @ 120+1.2 th 1 VLTC: Test final values (B) (Just a fast test to make sure it scales well at VLTC)
19-06-09 Ala master diff
ELO: 19.87 +-1.8 (95%) LOS: 100.0%
Total: 40000 W: 7081 L: 4796 D: 28123
40000 @ 60+0.6 th 1 Regression/progression test against SF10 after " Remove depth condition for ttPv" of June 9th.
19-06-09 vdv ttPV diff
LLR: -0.48 (-2.94,2.94) [0.00,3.50]
Total: 12909 W: 1696 L: 1694 D: 9519
sprt @ 60+0.6 th 8 test removal for Elo gain with large hash and more threads.
19-06-09 vdv ttPV diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 9721 W: 1582 L: 1753 D: 6386
sprt @ 60+0.6 th 1 LTC: Cleaned removal of ttPv.
19-06-08 Voy qsDCprune diff
LLR: 2.95 (-2.94,2.94) [0.00,3.50]
Total: 128428 W: 22222 L: 21679 D: 84527
sprt @ 60+0.6 th 1 LTC: Don't prune DC in qSearch
19-06-07 sg pawnsTableShrink diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 41844 W: 7104 L: 7123 D: 27617
sprt @ 60+0.6 th 1 Retry doubling pawn table size but also shrink pawn entry size from 112 to 88 bytes (-21.4%). Test as specualtive LTC to get enough pawn cache pressure and compared with the same version (seems good but not enough) without shrinking. Low TP
19-06-07 sg cmh_fromTo diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 24034 W: 3971 L: 4080 D: 15983
sprt @ 60+0.6 th 1 Speculative LTC. Because the table has about 4.6 more effective history indices it needs more time to reach saturation. So perhaps a longer search can compensate enough for this (Low TP). Add also a color index to history (like in main history). This doubles the table size additionally (so in the whole its now 8 times bigger: +14 MB per thread).
19-06-07 Voy ttPV diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 46366 W: 7867 L: 7912 D: 30587
sprt @ 60+0.6 th 1 Not ttPV at cutnodes. I think this may do well at LTC as there will be more ttHits since we are using a bigger hash table. Low TP.
19-06-05 sg pawnsTable diff
LLR: -1.82 (-2.94,2.94) [0.00,4.00]
Total: 127902 W: 22078 L: 21755 D: 84069
sprt @ 60+0.6 th 1 Speculative LTC because this should have more effect at higher depths because of increasing pawn cache pressure. Double size of pawns table. (Low TP)
19-06-05 Voy mcpT2 diff
LLR: 2.95 (-2.94,2.94) [0.00,3.50]
Total: 102115 W: 17692 L: 17224 D: 67199
sprt @ 60+0.6 th 1 Spec. LTC for this 130k patch. Low TP
19-06-06 MJZ Pos-Progression diff
LLR: -0.51 (-2.94,2.94) [0.00,3.50]
Total: 18811 W: 2942 L: 2922 D: 12947
sprt @ 60+0.6 th 2 Return VALUE_DRAW if we are moving the same pieces for several moves. rule50 > 38, ply > 38. Try VLTC SPRT.
19-06-06 MJZ Pos-Progression4 diff
ELO: 2.07 +-5.7 (95%) LOS: 75.9%
Total: 4365 W: 694 L: 668 D: 3003
5000 @ 60+0.6 th 2 Return VALUE_DRAW if we are moving the same pieces for several moves. Progressive decrease from rule50 > 28, ply > 28
19-06-06 MJZ Pos-Progression diff
ELO: 2.87 +-6.2 (95%) LOS: 81.6%
Total: 3875 W: 649 L: 617 D: 2609
5000 @ 60+0.6 th 2 Return VALUE_DRAW if we are moving the same pieces for several moves. rule50 > 38, ply > 38 - new run after some multi-thread corrections.
19-06-05 MJZ Pos-Progression2 diff
ELO: -0.83 +-5.5 (95%) LOS: 38.5%
Total: 4613 W: 697 L: 708 D: 3208
5000 @ 120+1.2 th 1 Return VALUE_DRAW if we are moving the same pieces for several moves. rule50 > 34, ply > 34
19-06-06 MJZ Pos-Progression diff
ELO: -77.92 +-51.6 (95%) LOS: 0.1%
Total: 68 W: 6 L: 21 D: 41
5000 @ 60+0.6 th 2 Return VALUE_DRAW if we are moving the same pieces for several moves. rule50 > 38, ply > 38
19-06-05 MJZ Pos-Progression diff
ELO: -0.07 +-5.3 (95%) LOS: 48.9%
Total: 4823 W: 719 L: 720 D: 3384
5000 @ 120+1.2 th 1 Return VALUE_DRAW if we are moving the same pieces for several moves. Limited test at VLTC to check the idea.
19-06-06 31m simplify_passerk diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 69548 W: 11855 L: 11813 D: 45880
sprt @ 60+0.6 th 1 LTC: Use intermediate value (k += 5).
19-06-05 MJZ Pos-Progression3 diff
ELO: 2.85 +-5.4 (95%) LOS: 84.8%
Total: 5000 W: 819 L: 778 D: 3403
5000 @ 120+1.2 th 1 Return VALUE_DRAW if we are moving the same pieces for several moves. rule50 > 36, ply > 36
19-06-05 pro ps_weakunopposed4 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 46605 W: 7969 L: 7890 D: 30746
sprt @ 60+0.6 th 1 LTC: 10/8