Stockfish Testing Queue

Pending - 0 tests 0.0 hrs

None

Active - 1 tests

26-03-17 SC easyMove01 diff
LLR: -2.18 (-2.94,2.94) [-3.00,1.00]
Total: 17486 W: 3121 L: 3267 D: 11098
sprt @ 10+0.1 th 1 Consecutive easy moves, but compensate in slow mover even more.

Finished - 699 tests

26-03-17 SC easyMove02 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 4471 W: 709 L: 872 D: 2890
sprt @ 10+0.1 th 1 Take 2. Do not reset best move changes on a ponder hit and allow consecutive easy moves. Fix bench. Much higher slow mover.
25-03-17 SC easyMove01 diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 25487 W: 4412 L: 4611 D: 16464
sprt @ 10+0.1 th 1 Consecutive easy moves, but compensate in slow mover.
25-03-17 SC dynamicContempt diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 21558 W: 3958 L: 3991 D: 13609
sprt @ 10+0.1 th 1 Final attempt. Superimpose with easy moves.
25-03-17 SC easyMove02 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 4501 W: 697 L: 859 D: 2945
sprt @ 10+0.1 th 1 Take 2. Do not reset best move changes on a ponder hit and allow consecutive easy moves. Fix bench.
24-03-17 SC dynamicContempt diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11295 W: 2002 L: 2079 D: 7214
sprt @ 10+0.1 th 1 One another try: do not interfere with easy moves.
20-03-17 SC dynamicContempt diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 71069 W: 9446 L: 9330 D: 52293
sprt @ 60+0.6 th 1 Half tuned values, including slow mover. Ltc
19-03-17 SC slowMoverIncrease diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 64189 W: 11607 L: 11545 D: 41037
sprt @ 10+0.1 th 1 As a control: slowMover += 1
20-03-17 SC dynamicContempt diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 40216 W: 5343 L: 5328 D: 29545
sprt @ 60+0.6 th 1 Half change tuned values, tttbno slow mover. Ltc
19-03-17 SC dynamicContempt diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 40240 W: 7338 L: 7291 D: 25611
sprt @ 10+0.1 th 1 Tuned values, no slow mover.
19-03-17 SC slowMoverIncrease diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 41146 W: 7407 L: 7424 D: 26315
sprt @ 10+0.1 th 1 As a control: slowMover += 6
19-03-17 SC dynamicContempt diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 34181 W: 6366 L: 6094 D: 21721
sprt @ 10+0.1 th 1 Half tuned values, including slow mover.
19-03-17 SC dynamicContempt diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16544 W: 3015 L: 3069 D: 10460
sprt @ 10+0.1 th 1 Tuned values, including slow mover.
19-03-17 SC dynamicContempt diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 12421 W: 2336 L: 2158 D: 7927
sprt @ 10+0.1 th 1 Half change tuned values, no slow mover.
18-03-17 SC dynamicContemptTune diff
30201/50000 iterations
61527/100000 games played
100000 @ 30+0.3 th 1 My stable version is doing better than the previous one. Set it on hold and try to tune at average TC, framework is not that busy anyway.
19-03-17 SC 2ndTTmove diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 26339 W: 4740 L: 4753 D: 16846
sprt @ 10+0.1 th 1 Inspired by https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/4FW_5oWRcKQ Surprisingly, bench does not change although on 10% of the cases ss->staticEval == VALUE_NONE but bestValue is'nt.
17-03-17 SC dynamicContempt diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 121593 W: 21966 L: 21577 D: 78050
sprt @ 10+0.1 th 1 Is this version more stable?
14-03-17 SC dynamicContemptCrunchy diff
ELO: 0.68 +-1.0 (95%) LOS: 91.5%
Total: 242036 W: 60329 L: 59853 D: 121854
200000 @ 2+0.02 th 1 Try to debug a fishtest problem, see https://groups.google.com/forum/#!msg/fishcooking/C3KorRmYmWM/g6MJjhI3BgAJ
13-03-17 SC dynamicContempt diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 14600 W: 1866 L: 1937 D: 10797
sprt @ 60+0.6 th 1 Only positive contempt. How does it scale? LTC with prior -1 in case framework goes idle.
14-03-17 SC dynamicContempt diff
LLR: -4.71 (-2.94,2.94) [0.00,5.00]
Total: 23501 W: 4254 L: 4353 D: 14894
sprt @ 10+0.1 th 1 Try to scale with depth, so that we can properly order the moves before using dynamic contempt too much.
13-03-17 SC dynamicContempt diff
LLR: -3.87 (-2.94,2.94) [0.00,5.00]
Total: 58582 W: 10722 L: 10635 D: 37225
sprt @ 10+0.1 th 1 Dynamic contempt is somewhat messing up time management. Try to be move slower.
13-03-17 SC dynamicContempt diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 31362 W: 5729 L: 5720 D: 19913
sprt @ 10+0.1 th 1 Or move faster.
12-03-17 SC dynamicContempt diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 53396 W: 9780 L: 9677 D: 33939
sprt @ 10+0.1 th 1 Only positive contempt
12-03-17 SC dynamicContempt diff
LLR: -4.15 (-2.94,2.94) [0.00,5.00]
Total: 31410 W: 5586 L: 5629 D: 20195
sprt @ 10+0.1 th 1 Only negative contempt
11-03-17 SC dynamicContempt diff
LLR: -3.00 (-2.94,2.94) [0.00,5.00]
Total: 40283 W: 7349 L: 7304 D: 25630
sprt @ 10+0.1 th 1 Take 3. Pushed to repo.
11-03-17 SC dynamicContempt diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 33241 W: 6072 L: 6055 D: 21114
sprt @ 10+0.1 th 1 Take 2.
11-03-17 SC moreCapturesScoring diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 3915 W: 642 L: 750 D: 2523
sprt @ 10+0.1 th 1 Take 4
11-03-17 SC moreCapturesScoring diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2075 W: 327 L: 444 D: 1304
sprt @ 10+0.1 th 1 Take 3.
09-03-17 SC dynamicContemptTune diff
44037/50000 iterations
89044/100000 games played
100000 @ 10+0.1 th 1 First attempt was not that bad. Tune! Change Pars.
10-03-17 SC moreCapturesScoring diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2720 W: 440 L: 554 D: 1726
sprt @ 10+0.1 th 1 Only use rank based capture scoring in capture_init
10-03-17 SC moreCapturesScoring diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 1858 W: 295 L: 413 D: 1150
sprt @ 10+0.1 th 1 Other way around
07-03-17 SC dynamicContemptTune diff
16802/50000 iterations
34131/100000 games played
100000 @ 10+0.1 th 1 First attempt was not that bad. Tune!
08-03-17 SC pawnAsy diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7253 W: 1247 L: 1341 D: 4665
sprt @ 10+0.1 th 1 A pawn idea
08-03-17 SC pawnAsy diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13104 W: 2294 L: 2364 D: 8446
sprt @ 10+0.1 th 1 A conservative version
06-03-17 SC ri01 diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 37645 W: 6516 L: 6735 D: 24394
sprt @ 10+0.1 th 1 Less reductions (half throughput)
06-03-17 SC dynamicContempt diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 27594 W: 5005 L: 5012 D: 17577
sprt @ 10+0.1 th 1 Dynamic contempt based on best move changes?
06-03-17 SC noAPP diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 28593 W: 4911 L: 5114 D: 18568
sprt @ 10+0.1 th 1 Was it a lucky run? Correct bounds.
06-03-17 SC ri01 diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 9996 W: 1695 L: 1868 D: 6433
sprt @ 10+0.1 th 1 More reductions (half throughput)
05-03-17 SC retireImproving diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 30238 W: 5260 L: 5467 D: 19511
sprt @ 10+0.1 th 1 Take 3.
05-03-17 SC impImp diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 4333 W: 705 L: 847 D: 2781
sprt @ 10+0.1 th 1 Does it make sense to calculate improving based on ss-1?
05-03-17 SC retireImproving diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 29373 W: 5145 L: 5351 D: 18877
sprt @ 10+0.1 th 1 Compensate by increasing reductions.
04-03-17 SC master diff
ELO: 89.31 +-3.4 (95%) LOS: 100.0%
Total: 12002 W: 3479 L: 460 D: 8063
40000 @ 30+0.3 th 1 Follow-up of discussion on forum https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/4SMIDpj2ma4 Very low throughput
04-03-17 SC master diff
ELO: 112.23 +-4.5 (95%) LOS: 100.0%
Total: 8977 W: 3269 L: 466 D: 5242
40000 @ 30+0.3 th 1 Follow-up of discussion on forum https://groups.google.com/forum/?fromgroups=#!topic/fishcooking/4SMIDpj2ma4 Very low throughput
04-03-17 SC retireImproving diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 32964 W: 5698 L: 5909 D: 21357
sprt @ 10+0.1 th 1 Do we really need improving in LMR?
04-03-17 SC retireImproving diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 15838 W: 2736 L: 2919 D: 10183
sprt @ 10+0.1 th 1 Or even more aggressive
24-02-17 SC historyEarlyPruning diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 31030 W: 3959 L: 3977 D: 23094
sprt @ 60+0.6 th 1 A related idea (two steps LMR). Speculative Ltc to know whether it makes sense to work on this. Prior low.
03-03-17 SC historyEarlyPruning diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 17820 W: 3051 L: 3102 D: 11667
sprt @ 10+0.1 th 1 This should be another small progress.
02-03-17 SC historyEarlyPruning diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 65594 W: 11523 L: 11375 D: 42696
sprt @ 10+0.1 th 1 Do it only for negative history
01-03-17 SC historyEarlyPruning diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 30079 W: 5375 L: 5373 D: 19331
sprt @ 10+0.1 th 1 Another attempt. (Sorry for tuning in this primitive way, but I think this would be really tough with SPSA).
01-03-17 SC historyEarlyPruning diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 7942 W: 1379 L: 1471 D: 5092
sprt @ 10+0.1 th 1 Start earlier.
26-02-17 SC historyEarlyPruning diff
LLR: -4.83 (-2.94,2.94) [0.00,5.00]
Total: 196967 W: 35175 L: 34553 D: 127239
sprt @ 10+0.1 th 1 Quadratic interpolation