Stockfish Testing Queue

Finished - 24165 tests

08-01-14 gl kingpp diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 4078 W: 718 L: 813 D: 2547
sprt @ 15+0.05 th 1 6 * enemy, 2 * friendly
07-01-14 ur union_of_patchs diff
LLR: 0.09 (-2.94,2.94) [-1.50,4.50]
Total: 3000 W: 549 L: 538 D: 1913
sprt @ 15+0.05 th 1 union of patchs version 2 that is based on my patch that passed also long time control(the queue is almost empty so I give it with a low priority).
07-01-14 ur time_branching_factor diff
ELO: 4.33 +-2.8 (95%) LOS: 99.9%
Total: 20000 W: 3480 L: 3231 D: 13289
20000 @ 40/30 th 1 I decided to test at this time control with fixed number of games the patch that passed stage 1 and stage 2(my plan is to add 20,000 more games unless I get less than 50%)
07-01-14 jo game_phase diff
ELO: -3.65 +-2.6 (95%) LOS: 0.3%
Total: 30000 W: 6388 L: 6703 D: 16909
30000 @ 5+0.05 th 1 Limits 3.
07-01-14 jo game_phase^^ diff
ELO: -7.30 +-2.6 (95%) LOS: 0.0%
Total: 30000 W: 6437 L: 7067 D: 16496
30000 @ 5+0.05 th 1 Calculating game_phase including pawns. Limits 1. Trying to find out, if the idea has potential.
05-01-14 mc fut_base diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 21434 W: 3437 L: 3439 D: 14558
sprt @ 60+0.05 th 1 LTC: FutilityBase tuning, final take!
07-01-14 jo game_phase^ diff
ELO: -4.23 +-2.6 (95%) LOS: 0.1%
Total: 30000 W: 6361 L: 6726 D: 16913
30000 @ 5+0.05 th 1 Limits 2.
07-01-14 sg lmr1 diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 12740 W: 2376 L: 2448 D: 7916
sprt @ 15+0.05 th 1 LMR: reduce losing checks more (equals here givesCheck && !ext)
07-01-14 in time_branching_factor diff
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 18019 W: 2936 L: 2752 D: 12331
sprt @ 60+0.05 th 1 LTC for ur: another try to use the branching factor to decide if to stop earlier and this time I try combination of the 0.62 patch that took a long time with not stopping earlier in the first move unless I already used some significant time in it without fail low
05-01-14 rs easy_move_byebye diff
ELO: 0.17 +-1.9 (95%) LOS: 57.0%
Total: 40000 W: 6397 L: 6377 D: 27226
40000 @ 60+0.05 th 1 remove the easy move code and add the condition to play instantly if only one legal move is available
07-01-14 ur union_of_patchs diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 11946 W: 2355 L: 2217 D: 7374
sprt @ 15+0.05 th 1 I try removing the easy move code together with some other changes that got yellow score and all of them do not make the code more complex.
05-01-14 ur faster_first_move diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 75062 W: 11907 L: 11665 D: 51490
sprt @ 60+0.05 th 1 test that passed short time control
06-01-14 tr KPST11 diff
ELO: -4.38 +-3.2 (95%) LOS: 0.4%
Total: 20000 W: 4378 L: 4630 D: 10992
20000 @ 5+0.05 th 1 KPST11: EG/2
06-01-14 ur time_branching_factor diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 26030 W: 4959 L: 4785 D: 16286
sprt @ 15+0.05 th 1 another try to use the branching factor to decide if to stop earlier and this time I try combination of the 0.62 patch that took a long time with not stopping earlier in the first move unless I already used some significant time in it without fail low
06-01-14 tr KPST14 diff
ELO: -3.68 +-3.2 (95%) LOS: 1.3%
Total: 20000 W: 4412 L: 4624 D: 10964
20000 @ 5+0.05 th 1 KPST14: MG/2
06-01-14 tr KPST12 diff
ELO: -23.38 +-3.4 (95%) LOS: 0.0%
Total: 19022 W: 3868 L: 5146 D: 10008
20001 @ 5+0.05 th 1 KPST12: EG*2
07-01-14 lb kpsqt diff
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 5890 W: 985 L: 857 D: 4048
sprt @ 60+0.05 th 1 LTC for Joona: KPSQT, try 1
06-01-14 jk kpsqt diff
ELO: 3.11 +-2.0 (95%) LOS: 99.9%
Total: 49665 W: 10941 L: 10497 D: 28227
50000 @ 5+0.05 th 1 KPSQT, try 1
07-01-14 lb kpsqt diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 16875 W: 3198 L: 3049 D: 10628
sprt @ 15+0.05 th 1 STC for Joona: KPSQT, try 1
06-01-14 tr KPST13 diff
ELO: -2.36 +-3.2 (95%) LOS: 7.7%
Total: 20000 W: 4492 L: 4628 D: 10880
20000 @ 5+0.05 th 1 KPST13: MG*2
06-01-14 lb 8d49f5c275ac0142da2f480 diff
LLR: -2.99 (-2.94,2.94) [-1.50,4.50]
Total: 22409 W: 5106 L: 5150 D: 12153
sprt @ 5+0.02 th 1 Piece values -20 Redo the test properly, setting Emergency Base Time on both test and base. Use asymmetric to reduce type I error.
06-01-14 rs se_random diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 2413 W: 413 L: 512 D: 1488
sprt @ 15+0.05 th 1 Take 2. Extend 65%
06-01-14 rs se_random diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 2626 W: 456 L: 555 D: 1615
sprt @ 15+0.05 th 1 Remove SE exclusion search. Instead randomly extend some of the TT moves. Take 1 extend 50%
06-01-14 lb 8d49f5c275ac0142da2f480 diff
ELO: 2.82 +-2.2 (95%) LOS: 99.5%
Total: 45000 W: 10495 L: 10130 D: 24375
40000 @ 5+0.02 th 1 Piece values -20
06-01-14 lb bbd1556d72b0df829ce716c diff
ELO: 0.05 +-2.3 (95%) LOS: 51.8%
Total: 40000 W: 9056 L: 9050 D: 21894
40000 @ 5+0.02 th 1 Piece values +20
29-12-13 jk no_check diff
ELO: -16.17 +-2.2 (95%) LOS: 0.0%
Total: 31187 W: 4499 L: 5949 D: 20739
40000 @ 60+0.05 th 1 LTC: Measure value of qsearch checks at depth zero.
05-01-14 ur time_again diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 2020 W: 339 L: 440 D: 1241
sprt @ 15+0.05 th 1 using the branching factor in time management(time during first move divided by time in the end of the last iteration. fixed a potential bug of uninitialized variable in test that did not start
05-01-14 sg futility diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 10314 W: 1594 L: 1648 D: 7072
sprt @ 60+0.05 th 1 LTC: change qsearch futility pruning condition
05-01-14 sg futility diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 11652 W: 2287 L: 2150 D: 7215
sprt @ 15+0.05 th 1 change qsearch futility pruning condition
04-01-14 tr KPST10 diff
ELO: 0.38 +-3.2 (95%) LOS: 59.2%
Total: 20000 W: 4481 L: 4459 D: 11060
20000 @ 5+0.05 th 1 KPST10: MG-table with file-bonus only and set to zero for the starting square. EG-table with reduced file-bonus (like KPST9)
04-01-14 tr KPST9a diff
ELO: -1.15 +-3.2 (95%) LOS: 24.2%
Total: 20000 W: 4418 L: 4484 D: 11098
20000 @ 5+0.05 th 1 KPST9a: Retest against current master (after retire_king_ring) with faster time control.
04-01-14 ur faster_first_move diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 48751 W: 9335 L: 9100 D: 30316
sprt @ 15+0.05 th 1 The long time control does not pass fast so I try with different parameters with the idea that the stop time is the same in the beginning of the iteration and in the first move if the move does not fail low
05-01-14 jo pp_npm diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 4310 W: 767 L: 861 D: 2682
sprt @ 15+0.05 th 1 Change the bonus for PP from less pawns into more non_pawn_material
03-01-14 ur faster_first_move diff
LLR: -2.86 (-2.94,2.94) [0.00,6.00]
Total: 56741 W: 9026 L: 8864 D: 38851
sprt @ 60+0.05 th 1 The last change of time management tweak with 0.62 instead of 0.8 that means stopping even faster during the first move
05-01-14 jo fut_base diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 5566 W: 1096 L: 976 D: 3494
sprt @ 15+0.05 th 1 FutilityBase tuning, final take!
05-01-14 lb interpolate diff
LLR: -1.31 (-2.94,2.94) [-3.00,3.00]
Total: 47287 W: 8998 L: 9045 D: 29244
sprt @ 15+0.05 th 1 Simplification: remove GrainSize
04-01-14 jo fut_base diff
ELO: -0.39 +-2.6 (95%) LOS: 38.3%
Total: 30000 W: 6470 L: 6504 D: 17026
30000 @ 5+0.05 th 1 Value(80)
04-01-14 rs lmr_turbo diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 4419 W: 782 L: 876 D: 2761
sprt @ 15+0.05 th 1 Dynamic late move reduction based on pilot search
04-01-14 jo fut_base^^^ diff
ELO: 1.02 +-2.6 (95%) LOS: 78.1%
Total: 30000 W: 6486 L: 6398 D: 17116
30000 @ 5+0.05 th 1 Value(96)
04-01-14 jo fut_base^ diff
ELO: 1.68 +-2.6 (95%) LOS: 89.9%
Total: 30000 W: 6517 L: 6372 D: 17111
30000 @ 5+0.05 th 1 Value(112). Fixed number of games at very short TC. Let's see how fast they finish and if they give reasonable results.
03-01-14 jk undef diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 9748 W: 1804 L: 1884 D: 6060
sprt @ 15+0.05 th 1 1 * popcount<Max15>(undefended)
03-01-14 jk undef diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 17116 W: 3222 L: 3282 D: 10612
sprt @ 15+0.05 th 1 2 * popcount<Max15>(undefended)
03-01-14 jk undef diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 38885 W: 7347 L: 7350 D: 24188
sprt @ 15+0.05 th 1 4 * popcount<Max15>(undefended)
03-01-14 st bishop_tweak diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 11554 W: 2180 L: 2255 D: 7119
sprt @ 15+0.05 th 1 Double the penalty for bishops in middle game phase for their starting position
03-01-14 fw easy_move_3 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 14047 W: 2633 L: 2701 D: 8713
sprt @ 15+0.05 th 1 variation on best version easy_move_2 (easy move take 5 - b corrected)
03-01-14 rs lmr_retreat diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 7416 W: 1349 L: 1435 D: 4632
sprt @ 15+0.05 th 1 Take 2.
03-01-14 rs lmr_retreat diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 30128 W: 5682 L: 5708 D: 18738
sprt @ 15+0.05 th 1 more LMR for retreats
03-01-14 lb no_rook_pst diff
LLR: -2.96 (-2.94,2.94) [-3.00,3.00]
Total: 13297 W: 2492 L: 2598 D: 8207
sprt @ 15+0.05 th 1 Remove Rook PST
02-01-14 gl master diff
ELO: 25.36 +-2.0 (95%) LOS: 100.0%
Total: 40000 W: 8286 L: 5371 D: 26343
40000 @ 60+0.05 th 1 Regression test after scaling eval with one pawn
03-01-14 ur time_manag_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,6.00]
Total: 36622 W: 5837 L: 5770 D: 25015
sprt @ 60+0.05 th 1 reduce the time to play during first move by 20%. I put it with a lower priority because I prefer first to test another test with fast time control.