Stockfish Testing Queue

Finished - 20666 tests

25-04-14 My 3pawns_vs_minor diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 21002 W: 3678 L: 3729 D: 13595
sprt @ 15+0.05 th 1 Favour 3 pawns over a minor
24-04-14 in closed_position_minors diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 9555 W: 1720 L: 1800 D: 6035
sprt @ 15+0.05 th 1 Take 2: Use large array instead of formula.
24-04-14 ra rook_vs_minors_C diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 33711 W: 5931 L: 5950 D: 21830
sprt @ 15+0.05 th 1 Add rook vs minors rules. Take 3.
24-04-14 ra rook_vs_minors_B diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 25098 W: 4429 L: 4470 D: 16199
sprt @ 15+0.05 th 1 Add rook vs minors rules. Take 2.
24-04-14 in stormdanger diff
LLR: -2.97 (-2.94,2.94) [0.00,6.00]
Total: 16266 W: 2463 L: 2492 D: 11311
sprt @ 60+0.05 th 1 LTC: Take 3
24-04-14 Ro tt_thread_safe diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 1409 W: 163 L: 260 D: 986
sprt @ 15+0.05 th 3 Thread safe TT. Fixed a bug in TT:store. Requires multiple threads to test.
24-04-14 ra rook_vs_minors diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 16883 W: 2460 L: 2487 D: 11936
sprt @ 60+0.05 th 1 LTC: Add some rook vs minor(s) knowledge. Fixed version.
24-04-14 in stormdanger diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 11872 W: 2299 L: 2162 D: 7411
sprt @ 15+0.05 th 1 STC: Take 3
23-04-14 in stormdanger diff
ELO: 1.65 +-2.0 (95%) LOS: 95.1%
Total: 50000 W: 10418 L: 10181 D: 29401
50000 @ 5+0.05 th 1 Final Take: 75K iterations. Best take goes to SPRT. It will be hard to best Take 3 :)
21-04-14 in stormdanger diff
ELO: 1.38 +-2.0 (95%) LOS: 91.4%
Total: 50000 W: 10569 L: 10371 D: 29060
50000 @ 5+0.05 th 1 Take 1: 25K iterations. (My last tuning attempt was a disaster because I messed up the tuning set up. Hopefully, this time, it is much better.)
23-04-14 in stormdanger diff
ELO: 3.54 +-2.0 (95%) LOS: 100.0%
Total: 50000 W: 10592 L: 10082 D: 29326
50000 @ 5+0.05 th 1 Take 3 (I have attempted to correct what I believe is a small bug in the old logic -- read commit description)
24-04-14 ra rook_vs_minors diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 9157 W: 1678 L: 1551 D: 5928
sprt @ 15+0.05 th 1 Add some rook vs minor(s) knowledge. Fixed version.
24-04-14 ra king_safety2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 17060 W: 2963 L: 3024 D: 11073
sprt @ 15+0.05 th 1 Reduce king shelter penalty when enemy king is close. Take 2.
24-04-14 Ro tt_thread_safe diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 2071 W: 271 L: 367 D: 1433
sprt @ 15+0.05 th 3 Thread-safe TT. Requires running on multiple threads to test.
23-04-14 pe contempt diff
LLR: 2.99 (-2.94,2.94) [-1.50,4.50]
Total: 68650 W: 12448 L: 12168 D: 44034
sprt @ 15+0.05 th 1 One more STC test. Combination of two values which seems best on increase and decrease of eval.
24-04-14 in closed_position_storm diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 10743 W: 1876 L: 1953 D: 6914
sprt @ 15+0.05 th 1 Take 2: Different approach.
24-04-14 in closed_position_minors diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 15917 W: 2765 L: 2829 D: 10323
sprt @ 15+0.05 th 1 Simple bonus and malus for minor pieces in closed position (partially tuned).
23-04-14 in closed_position_storm diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 9854 W: 1720 L: 1799 D: 6335
sprt @ 15+0.05 th 1 Pawn storms are more dangerous when the position is locked at the center.
23-04-14 in rookon7th_simplify diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 9238 W: 1448 L: 1311 D: 6479
sprt @ 60+0.05 th 1 LTC: Remove RookOn7th and merge mg values into psqt (functional change)
23-04-14 pe contempt diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 45365 W: 8349 L: 8128 D: 28888
sprt @ 15+0.05 th 1 Double decrease of contempt when eval is negative
23-04-14 ra rook_vs_minors diff
LLR: 0.15 (-2.94,2.94) [-1.50,4.50]
Total: 344 W: 57 L: 51 D: 236
sprt @ 15+0.05 th 1 Penalize rook vs two minor, and add bonus for rook vs one minor. Idea from Lyudmil Tsvetkov at talkchess.
21-04-14 in rook_passers diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 29468 W: 5431 L: 5325 D: 18712
sprt @ 15+0.05 th 1 Simplify rook passers (Low priority)
21-04-14 in minor_threats diff
ELO: -8.46 +-2.5 (95%) LOS: 0.0%
Total: 30000 W: 5789 L: 6519 D: 17692
30000 @ 5+0.05 th 1 Calculate elo impact of non-pawn threat stuff only.
20-04-14 ra king_safety diff
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 50619 W: 7506 L: 7387 D: 35726
sprt @ 60+0.05 th 1 LTC: Penalize bad king pawns if enemy king is in close file.
23-04-14 pe contempt diff
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 12674 W: 2322 L: 2186 D: 8166
sprt @ 15+0.05 th 1 Decrease eval-based contempt for negative eval
22-04-14 in passers_pawncount diff
LLR: -3.14 (-2.94,2.94) [-3.00,1.00]
Total: 27680 W: 4904 L: 5118 D: 17658
sprt @ 15+0.05 th 1 Remove pawn count in passed pawn eval.
19-04-14 mc knight_pawns diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 67474 W: 9974 L: 9921 D: 47579
sprt @ 60+0.05 th 1 LTC as simplification: Remove penalty for knights when there are few enemy pawns. (This qualifies as a simplification, but I think there is some gain in elo too)
22-04-14 in rook_passers3 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 17727 W: 3248 L: 3122 D: 11357
sprt @ 15+0.05 th 1 Remove rook passers (test as simplification)
22-04-14 lb contempt diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 92941 W: 17175 L: 17179 D: 58587
sprt @ 15+0.05 th 1 use fixd contempt
23-04-14 jh illogical_pawns diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 5917 W: 1069 L: 1159 D: 3689
sprt @ 15+0.05 th 1 Remove illogical pawn condition.
22-04-14 pe contempt diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 31133 W: 4761 L: 4723 D: 21649
sprt @ 60+0.05 th 1 LTC. Double eval based contempt
22-04-14 jo clop_qimb diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 32413 W: 4967 L: 4864 D: 22582
sprt @ 60+0.05 th 1 LTC: Replace my queen vs. 3 minors rule by CLOP values. Test in no regression mode.
21-04-14 mc master diff
ELO: 54.53 +-2.1 (95%) LOS: 100.0%
Total: 40000 W: 10969 L: 4742 D: 24289
40000 @ 60+0.05 th 1 Regression test after contempt (test contempt vs a weaker engine)
22-04-14 lb rookon7th_simplify diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 55678 W: 9954 L: 9892 D: 35832
sprt @ 15+0.05 th 1 Reschedule with correct SPRT parameters for Arjun: Remove RookOn7th and merge mg values into psqt (functional change)
21-04-14 pe time_contempt diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 27772 W: 5035 L: 5068 D: 17669
sprt @ 15+0.05 th 1 Simple addition of time contempt (raise contempt if more time ) on top of eval based, to see if time contempt adds something extra.
22-04-14 jo clop_qimb diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 17254 W: 3159 L: 3032 D: 11063
sprt @ 15+0.05 th 1 Replace my queen vs. 3 minors rule by CLOP values. Test in no regression mode.
22-04-14 in stormdanger diff
ELO: 1.96 +-2.0 (95%) LOS: 97.5%
Total: 50000 W: 10558 L: 10276 D: 29166
50000 @ 5+0.05 th 1 Take 2: 42K iterations
22-04-14 pe contempt diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 16642 W: 3029 L: 3091 D: 10522
sprt @ 15+0.05 th 1 Triple eval based contempt. This is almost Mysseno's value ~37 in originally passed test, though it was symmetric -37/37.
21-04-14 pe contempt diff
LLR: 2.95 (-2.94,2.94) [-1.50,4.50]
Total: 22133 W: 4084 L: 3923 D: 14126
sprt @ 15+0.05 th 1 Double eval based contempt
21-04-14 in threat_eval diff
ELO: -9.07 +-2.5 (95%) LOS: 0.0%
Total: 30000 W: 5775 L: 6558 D: 17667
30000 @ 5+0.05 th 1 Calculate elo impact of threat stuff.
21-04-14 in spacetrade diff
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 14776 W: 2174 L: 2210 D: 10392
sprt @ 60+0.05 th 1 Take 3: Scaling the penalty based on number of pieces attacked.
21-04-14 My weakPoint diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 4481 W: 765 L: 858 D: 2858
sprt @ 15+0.05 th 1 Bonus if enemy has weaknesses
21-04-14 pe contempt diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 23744 W: 4163 L: 4207 D: 15374
sprt @ 15+0.05 th 1 halve eval based contempt
21-04-14 mc rook_passers diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 39829 W: 7138 L: 7141 D: 25550
sprt @ 15+0.05 th 1 Tweak rook passers formula according to Lyudmil suggestion on talkchess: last take
21-04-14 mc rook_passers^ diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 5263 W: 936 L: 1027 D: 3300
sprt @ 15+0.05 th 1 Tweak rook passers formula according to Lyudmil suggestion on talkchess: take 3
21-04-14 mc rook_passers diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 16806 W: 3082 L: 3143 D: 10581
sprt @ 15+0.05 th 1 Tweak rook passers formula according to Lyudmil suggestion on talkchess: take 2
20-04-14 mc storm_respin diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 42693 W: 7777 L: 7772 D: 27144
sprt @ 15+0.05 th 1 Another attempt at pawn storm tweak
20-04-14 in spacetrade diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 43257 W: 7630 L: 7419 D: 28208
sprt @ 15+0.05 th 1 Take 3: Scaling the penalty based on number of pieces attacked.
19-04-14 in trapped_rook_simplify diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 49465 W: 8522 L: 8760 D: 32183
sprt @ 15+0.05 th 1 Simplification: Checking for semi-open file is redundant.
19-04-14 gl storm_retest diff
ELO: 1.81 +-1.9 (95%) LOS: 97.2%
Total: 40000 W: 6039 L: 5831 D: 28130
40000 @ 60+0.05 th 1 Measure elo gain from storm patch that passed LTC SPRT 0..6