Stockfish Testing Queue

Finished - 37328 tests

15-10-18 SC assorted_tuning diff
LLR: 3.07 (-2.94,2.94) [0.00,4.00]
Total: 15124 W: 2974 L: 2756 D: 9394
sprt @ 10+0.1 th 1 Retest assorted tuning without time management as discussed in https://github.com/official-stockfish/Stockfish/pull/464
15-10-17 sni king_march diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 33232 W: 6508 L: 6485 D: 20239
sprt @ 10+0.1 th 1 Try to use vertical king separation info
15-10-18 mco lazy_smp diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 3607 W: 674 L: 526 D: 2407
sprt @ 10+0.1 th 7 lazy smp pre-merge test: middle case scenario. With 7 threads YBW is still godd but lazy should start to be effective. Test as simplification because lazy_smp saves 385(!) less lines of code.
15-10-17 aji hybrid_history diff
ELO: 2.92 +-3.9 (95%) LOS: 93.0%
Total: 10000 W: 1669 L: 1585 D: 6746
10000 @ 10+0.1 th 7 Use a shared history/countermoves within the cluster: STC
15-10-18 sni pawn_asymmetry5 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 12504 W: 2322 L: 2436 D: 7746
sprt @ 10+0.1 th 1 Double the asymmetry bonus
15-10-17 Roc InitiativeBlocked diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 25788 W: 5055 L: 5066 D: 15667
sprt @ 10+0.1 th 1 IB_20151017_1
15-10-17 mco lazy_smp diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 40232 W: 6908 L: 7130 D: 26194
sprt @ 10+0.1 th 3 lazy smp pre-merge test: worst case scenario. With 3 threads YBW scales almost in an ideal way so this is a tough test for lazy. Test as simplification becasue lazy_smp saves 385(!) less lines of code.
15-10-17 sni blocked_pawns2 diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 24640 W: 4687 L: 4704 D: 15249
sprt @ 10+0.1 th 1 Take 2 (half malus for blocked pawns)
15-10-17 mco lazy_smp diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 28186 W: 3908 L: 3798 D: 20480
sprt @ 60+0.1 th 3 lazy smp pre-merge test: worst case scenario. With 3 threads YBW scales almost in an ideal way so this is a tough test for lazy. Test as simplification becasue lazy_smp saves 385(!) less lines of code. LTC version.
15-10-17 sni threats7 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 33400 W: 6402 L: 6441 D: 20557
sprt @ 10+0.1 th 1 Values of threats on attacked queens, take 4 (change=S(8,8)). Rebased on current master.
15-10-17 sni rook5 diff
LLR: -2.79 (-2.94,2.94) [0.00,5.00]
Total: 19292 W: 3668 L: 3702 D: 11922
sprt @ 10+0.1 th 1 Also simplify logic in SF's KRPPKRP endgame function
15-10-17 SC initiative_pawns diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 18987 W: 3604 L: 3647 D: 11736
sprt @ 10+0.1 th 1 Try the opposite.
15-10-17 sg initiative diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 6555 W: 1213 L: 1311 D: 4031
sprt @ 10+0.1 th 1 Use stronger side passed pawns for initiative.
15-10-17 Roc InitiativeMg diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16166 W: 3031 L: 3086 D: 10049
sprt @ 10+0.1 th 1 Using mg=eg/2 instead of 0, and max=4 on king_separation
15-10-17 sg initiative diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10048 W: 1926 L: 2008 D: 6114
sprt @ 10+0.1 th 1 Use weaker side passed pawns for initiative.
15-10-17 sni blocked_pawns2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10912 W: 2052 L: 2130 D: 6730
sprt @ 10+0.1 th 1 Blocked pawn malus for the attacking side
15-10-17 SC initiative_pawns diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8467 W: 1596 L: 1685 D: 5186
sprt @ 10+0.1 th 1 Only use pawn count from strong side in scaling endgame.
15-10-17 lbr test diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 1159 W: 144 L: 263 D: 752
sprt @ 10+0.1 th 3 anarchic smp
15-10-16 IIv time_management diff
LLR: -1.24 (-2.94,2.94) [-3.00,1.00]
Total: 62543 W: 12016 L: 12195 D: 38332
sprt @ 10+0.1 th 1 Using 75% of time and the whole increment should cover most needs, both in making good moves, and leaving enough time on the clock. It seems like a natural limitation to me.
15-10-16 aji hybrid1 diff
ELO: -8.74 +-11.2 (95%) LOS: 6.3%
Total: 1193 W: 178 L: 208 D: 807
10000 @ 10+0.1 th 20 hybrid with 20 threads: STC (hardcoded Min Split Depth to 5)
15-10-15 sg update_stats diff
LLR: -3.76 (-2.94,2.94) [0.00,4.00]
Total: 58979 W: 11295 L: 11287 D: 36397
sprt @ 10+0.1 th 1 Test first the tuned history update. But possible further tuning is necessary (some values seems not converged).
15-10-16 sg update_stats diff
LLR: -3.10 (-2.94,2.94) [0.00,4.00]
Total: 83734 W: 16097 L: 15965 D: 51672
sprt @ 10+0.1 th 1 Test tuned values. Take 2
15-10-17 Roc KingSeparationVerificat diff
LLR: -3.36 (-2.94,2.94) [-3.00,1.00]
Total: 35815 W: 6877 L: 7123 D: 21815
sprt @ 10+0.1 th 1 Making sure we are building on a solid ground here, before adding more ideas.
15-10-17 lbr min_thinking_time diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 3968 W: 734 L: 903 D: 2331
sprt @ 15+0 th 1 take 2. no increment.
15-10-16 lbr assorted_tuning diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 31284 W: 6052 L: 6039 D: 19193
sprt @ 10+0.1 th 1 Stefano80's combo patch, without its dubious parts.
15-10-16 sni threats6 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 51283 W: 9924 L: 9822 D: 31537
sprt @ 10+0.1 th 1 Values of threats on attacked queens, take 3 (change=S(9,9))
15-10-16 lbr min_thinking_time diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 53831 W: 10239 L: 10176 D: 33416
sprt @ 10+0.1 th 1 take 2
15-10-17 sni rook5 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16450 W: 3181 L: 3234 D: 10035
sprt @ 10+0.1 th 1 Gull's KPRRRKRPP : take 2
15-10-16 IIv tmm_tuning diff
9912/10000 iterations
20000/20000 games played
20000 @ 20+0.2 th 1 Tuning time management. 20K run to see in which direction this goes.
15-10-16 jhe simple_time_2 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 10524 W: 1962 L: 2140 D: 6422
sprt @ 10+0.1 th 1 Further simplification, closer emulation of old behavior.
15-10-16 Voy LMRt-b diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4072 W: 740 L: 849 D: 2483
sprt @ 10+0.1 th 1 LMR Tweak.
15-10-16 sg tune_history diff
19658/20000 iterations
40000/40000 games played
40000 @ 20+0.2 th 1 SPRT test struggles, so tune history update further from last values. Lower ck values for decay and weight
15-10-16 bin krppkrp_mod diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16298 W: 3100 L: 3154 D: 10044
sprt @ 10+0.1 th 1 Even more conservative and continuous
15-10-16 sni rook5 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12085 W: 2286 L: 2359 D: 7440
sprt @ 10+0.1 th 1 Code Gull's rules for KRPPPKRPP endgame
15-10-16 Voy LMRt-a4 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 6823 W: 1253 L: 1350 D: 4220
sprt @ 10+0.1 th 1 Take 4: Try one more shot at this type of LMR tweak.
15-10-16 sni threats6 diff
LLR: -2.41 (-2.94,2.94) [0.00,4.00]
Total: 15298 W: 2939 L: 3014 D: 9345
sprt @ 10+0.1 th 1 Values of threats on attacked queens, take 4 (change=S(6,6))
15-10-16 aji hybrid diff
LLR: -0.02 (-2.94,2.94) [0.00,5.00]
Total: 15 W: 4 L: 5 D: 6
sprt @ 20+0.2 th 20 Try hybrid at 20 threads : Medium TC to exercise YBW more
15-10-16 bin krppkrp_mod diff
LLR: -0.29 (-2.94,2.94) [0.00,5.00]
Total: 14238 W: 2742 L: 2691 D: 8805
sprt @ 10+0.1 th 1 more conservative values for KRPPKRP will bring Elo?
15-10-16 Voy LMRt-a3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10400 W: 1942 L: 2023 D: 6435
sprt @ 10+0.1 th 1 Take 3
15-10-16 jos lazy_smp_cheng diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 2708 W: 426 L: 539 D: 1743
sprt @ 10+0.1 th 3 Only alter the search depth, if it is smaller than the main thread's one.
15-10-16 aji hybrid diff
ELO: -5.09 +-22.0 (95%) LOS: 32.5%
Total: 273 W: 37 L: 41 D: 195
10000 @ 10+0.1 th 20 Try hybrid at 20 threads: STC and set Min Split Depth to default value.
15-10-16 bin rook_eg diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 6558 W: 1227 L: 1325 D: 4006
sprt @ 10+0.1 th 1 Try 4, All pawn together + conservative scales
15-10-16 bin rook_eg diff
LLR: -3.22 (-2.94,2.94) [0.00,5.00]
Total: 10197 W: 1913 L: 2006 D: 6278
sprt @ 10+0.1 th 1 Try 3, Scale rook endgame now with linked pawns condition
15-10-16 lbr min_thinking_time diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 12568 W: 2371 L: 2553 D: 7644
sprt @ 15+0 th 1 take 1. without increment.
15-10-16 SC assorted_tuning diff
LLR: 3.01 (-2.94,2.94) [0.00,5.00]
Total: 13886 W: 2239 L: 2063 D: 9584
sprt @ 40+0.4 th 1 There was some almost passed tuning attempts in the last month. Collect them and give them a second chance. LTC.
15-10-16 Mys RT diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10200 W: 1907 L: 1989 D: 6304
sprt @ 10+0.1 th 1 Try adjusting the scoring, take 2.
15-10-16 bin rook_eg diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10352 W: 2004 L: 2085 D: 6263
sprt @ 10+0.1 th 1 Scale drawn rook endgames with king invasion check
15-10-16 Mys RT diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18249 W: 3540 L: 3585 D: 11124
sprt @ 10+0.1 th 1 Try this ranked threats business
15-10-13 lbr atomic_signals diff
ELO: 0.64 +-2.6 (95%) LOS: 68.3%
Total: 20000 W: 3053 L: 3016 D: 13931
20000 @ 20+0.05 th 15 Test at long tc, with many threads, suggested by Joona
15-10-15 lbr min_thinking_time diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 29907 W: 5637 L: 5532 D: 18738
sprt @ 10+0.1 th 1 Attempt to correct minimum thinking time logic. Apply as a minimum, instead of an add-on, and use a dynamic value, instead of a hardcoded one.