Stockfish Testing Queue

Finished - 4219 tests

19-04-24 pro ps_rookonfile2 diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 41026 W: 6899 L: 7121 D: 27006
sprt @ 60+0.6 th 1 LTC: simplify rookOnFile: 17,6
19-04-22 mco shuffle diff
ELO: -0.17 +-2.6 (95%) LOS: 44.8%
Total: 20000 W: 2890 L: 2900 D: 14210
20000 @ 180+1.8 th 1 Shuffle var 7: no stop rule VLTC
19-04-23 Viz LMRmcCond3 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 22008 W: 3654 L: 3768 D: 14586
sprt @ 60+0.6 th 1 Instead of trying different parameter values I want to see how it scales - if it fails as fast as previous LTC there is nothing there. Normalized TP.
19-04-23 pro ps_connected19 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 27780 W: 4682 L: 4572 D: 18526
sprt @ 60+0.6 th 1 LTC: simplify connected (each factor in it's own term).
19-04-23 pro ps_passed101 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 50732 W: 8582 L: 8510 D: 33640
sprt @ 60+0.6 th 1 LTC: try to simplify passed pawns.
19-04-23 Viz LMRmcCond1 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 21925 W: 3629 L: 3743 D: 14553
sprt @ 60+0.6 th 1 LTC for this one...
19-04-22 Viz LMRCondSS1 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 70981 W: 11956 L: 11932 D: 47093
sprt @ 60+0.6 th 1 LTC
19-04-22 xot initpassed1 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 18503 W: 3088 L: 3183 D: 12232
sprt @ 60+0.6 th 1 Higher weight for passed_count in initiative, 11.
19-04-22 mco shuffle diff
ELO: -1.51 +-3.3 (95%) LOS: 18.8%
Total: 11981 W: 1697 L: 1749 D: 8535
20000 @ 180+1.8 th 1 Shuffle var 4: no search VLTC
19-04-21 mco shuffle diff
ELO: 0.06 +-2.9 (95%) LOS: 51.7%
Total: 16149 W: 2361 L: 2358 D: 11430
20000 @ 180+1.8 th 1 Shuffle var 3: stop rule VLTC
19-04-21 mco shuffle^ diff
ELO: 0.21 +-2.8 (95%) LOS: 55.8%
Total: 19932 W: 3368 L: 3356 D: 13208
20000 @ 60+0.6 th 1 Shuffle var 2: search window
19-04-21 mco shuffle diff
ELO: 0.45 +-2.8 (95%) LOS: 62.5%
Total: 19872 W: 3321 L: 3295 D: 13256
20000 @ 60+0.6 th 1 Shuffle var 3: stop rule
19-04-21 mco shuffle^^ diff
ELO: -1.25 +-2.8 (95%) LOS: 19.3%
Total: 19973 W: 3405 L: 3477 D: 13091
20000 @ 60+0.6 th 1 Shuffle var 1: tte->depth
19-04-20 mco shuffle diff
ELO: -0.09 +-2.8 (95%) LOS: 47.6%
Total: 20000 W: 3323 L: 3328 D: 13349
20000 @ 60+0.6 th 1 Test of shuffle code. Just a quick test at LTC to avoid spending resources if something is very wrong.
19-04-21 31m 842692d5b906984b340beb9 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 14541 W: 2325 L: 2460 D: 9756
sprt @ 60+0.6 th 1 With more than half of our workers idle, speculative LTC to check scaling of the best of the many yellow tests on the KD_relrank branch. STC 61K yellow. Low throughput.
19-04-20 Viz SpaceWTw1 diff
LLR: -2.94 (-2.94,2.94) [0.00,4.00]
Total: 16683 W: 2761 L: 2861 D: 11061
sprt @ 60+0.6 th 1 LTC, normalized TP.
19-04-19 31m 670c970a3b923cbe4513050 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 30639 W: 5219 L: 5308 D: 20112
sprt @ 60+0.6 th 1 There's no reason to leave the framework empty. Speculative LTC for @mstembera's 125K STC yellow.
19-04-19 Viz NmpTweak4 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 24788 W: 4064 L: 4119 D: 16605
sprt @ 60+0.6 th 1 Quick and fast check of scaling for this patch - this is triggered exclusively in deep endgames so should be depth dependant. STC SPRT bounds for faster converging. Normalized TP.
19-04-17 Viz SideBishopPR1 diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 103775 W: 17603 L: 17486 D: 68686
sprt @ 60+0.6 th 1 Prio -1 in case network goes idle
19-04-18 vdv reductSF diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 93846 W: 15964 L: 15962 D: 61920
sprt @ 60+0.6 th 1 Remove capping in reduction.
19-04-18 Viz FutMargTweak6 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 11205 W: 1792 L: 1910 D: 7503
sprt @ 60+0.6 th 1 LTC
19-04-17 Viz SideBishopPR1 diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 34005 W: 5786 L: 5866 D: 22353
sprt @ 60+0.6 th 1 Rerun of LTC
19-04-16 MJZ Shuffle-Tune7 diff
LLR: 2.95 (-2.94,2.94) [0.00,3.50]
Total: 57835 W: 8633 L: 8316 D: 40886
sprt @ 180+1.8 th 1 Try to improve Shuffle-Tune6 - Extension limit = 18. STC non regression is OK, testing VLTC with low thp.
19-04-17 pro ps_piecelist2 diff
LLR: -0.56 (-2.94,2.94) [-3.00,1.00]
Total: 2822 W: 475 L: 509 D: 1838
sprt @ 60+0.6 th 1 LTC: a bit faster for square (don't need to pop here).
19-04-16 Viz SideBishop1 diff
LLR: 2.95 (-2.94,2.94) [0.00,3.50]
Total: 234796 W: 40052 L: 39212 D: 155532
sprt @ 60+0.6 th 1 Spec. LTC for the best attempt. Normalized TP.
19-04-16 pro ps_distancering1 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 35470 W: 6015 L: 5918 D: 23537
sprt @ 60+0.6 th 1 STC was suspicious. Testing LTC for verification. Very LOW tp, and prio: -1.
19-04-17 pro ps_piecelist2 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 49098 W: 8215 L: 8449 D: 32434
sprt @ 60+0.6 th 1 Ltc; is pop_lsb at least as fast as our pieceLists and index? (remove pieceList and index). Bench changes because of move ordering.
19-04-16 MJZ Shuffle-Tune7 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 72809 W: 12335 L: 12305 D: 48169
sprt @ 60+0.6 th 1 Try to improve Shuffle-Tune6 - Extension limit = 18. STC non regression is OK, testing LTC with low thp.
19-04-15 Cof novoting diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 13519 W: 1919 L: 2057 D: 9543
sprt @ 60+0.6 th 4 Prove that current master is bad with 4 threads and 60+0.6 time control
19-04-15 vdv earlyPrunePV diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 120001 W: 20263 L: 20303 D: 79435
sprt @ 60+0.6 th 1 Take 2. Combined.
19-04-15 pro ps_semiopen7 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 10173 W: 1774 L: 1636 D: 6763
sprt @ 60+0.6 th 1 LTC: try to simplify semiopen_files. ( / 4)
19-04-15 MJZ Shuffle-Tune6 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 88749 W: 15123 L: 15047 D: 58579
sprt @ 60+0.6 th 1 LTC test of http://tests.stockfishchess.org/tests/view/5cb10da60ebc5925cf013c74
19-04-13 MJZ Shuffle-Tune4 diff
LLR: -1.95 (-2.94,2.94) [0.00,3.50]
Total: 75707 W: 10926 L: 10851 D: 53930
sprt @ 180+1.8 th 1 Shuffle limit = 36 - 6 x (piece.count > 14), extension limited to 2 x rootDepth - Test at VLTC (please accept only if LTC is OK and then VLTC is not spec).
19-04-12 MJZ Shuffle-Tune6 diff
LLR: 2.95 (-2.94,2.94) [0.00,3.50]
Total: 67356 W: 9963 L: 9623 D: 47770
sprt @ 180+1.8 th 1 Back to more safe approach. Shuffle limit = 36 - 6 x (piece.count > 14)
19-04-14 svi cyclefix2 diff
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 19620 W: 3308 L: 3185 D: 13127
sprt @ 60+0.6 th 1 LTC: Fix cycle detection in presence of repetitions search() may incorrectly return a draw score in the following corner case: There was a 2-fold repetition during the game, and the current postion can be reached by a move from a repeated one. This case is treated as an upcoming 3-fold repetition, which it is not. Here is a testcase demonstrating the issue. (Note that the moves after FEN are required). The input position fen 8/8/8/8/8/8/p7/2k4K b - - 0 1 moves c1b1 h1g1 b1c1 g1h1 c1b1 h1g1 b1a1 g1h1 go movetime 1000 produces the output [...] info depth 127 seldepth 2 multipv 1 score cp 0 [...] bestmove a1b1 saying that the game will be drawn by repetion. However the other possible move for black, Kb2, avoids repetitions and wins. The patch fixes this behavior. In particular it finds mate in 10 in the above position.
19-04-14 vdv shuffleExtend diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 13168 W: 2195 L: 2296 D: 8677
sprt @ 60+0.6 th 1 Take 5. spec LTC
19-04-14 pro ps_connected102 diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 25539 W: 4223 L: 4419 D: 16897
sprt @ 60+0.6 th 1 try to simplify connected. v, v/2
19-04-13 MJZ Shuffle-Tune4 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 45883 W: 7633 L: 7680 D: 30570
sprt @ 60+0.6 th 1 Shuffle limit = 36 - 6 x (piece.count > 14), extension limited to 2 x rootDepth - spec LTC
19-04-14 Viz DiscoCheckKD2 diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 52350 W: 8781 L: 8810 D: 34759
sprt @ 60+0.6 th 1 LTC
19-04-13 vdv shuffleExtend^ diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 13849 W: 2290 L: 2427 D: 9132
sprt @ 60+0.6 th 1 Take 2. spec LTC.
19-04-12 MJZ Shuffle-Tune4 diff
LLR: -0.24 (-2.94,2.94) [0.00,3.50]
Total: 19735 W: 2834 L: 2799 D: 14102
sprt @ 180+1.8 th 1 Shuffle detection : ply < 3 * rootDepth - Framework is empty, I try at VLTC with low thp
19-03-26 Cof novoting diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 117824 W: 15660 L: 15966 D: 86198
sprt @ 60+0.6 th 8 novoting seems to be an elo gain on 60+0.6 4 threads, so test it on 60+0.6 8 threads, too. No functional change on 1 thread
19-04-12 Elb bwip3 diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 21100 W: 3480 L: 3597 D: 14023
sprt @ 60+0.6 th 1 Speculative LTC for http://tests.stockfishchess.org/tests/view/5cb042660ebc5925cf012155 It seems logical to me that the greater penalty (backward) should be given first.
19-04-12 MJZ Shuffle-Tune4 diff
LLR: -2.20 (-2.94,2.94) [0.00,3.50]
Total: 51503 W: 8609 L: 8596 D: 34298
sprt @ 60+0.6 th 1 Shuffle detection : ply < 3 * rootDepth, one more try with d < 3 - Spec LTC (needs high depth search)
19-04-12 MJZ Shuffle-Tune4 diff
LLR: -1.74 (-2.94,2.94) [0.00,3.50]
Total: 19875 W: 3300 L: 3348 D: 13227
sprt @ 60+0.6 th 1 After last patch removing dangerous shuffle procedure, test a new one with better parameters and limited loops to avoid infinite search - try 2 - ply < 3 * rootDepth - Spec LTC (needs high depth search)
19-04-12 MJZ Shuffle-Tune4 diff
LLR: -0.32 (-2.94,2.94) [0.00,3.50]
Total: 20975 W: 3606 L: 3565 D: 13804
sprt @ 60+0.6 th 1 Shuffle limit : decrease by 12 if pieces count > 14.
19-04-12 MJZ Shuffle-Tune5 diff
LLR: -0.62 (-2.94,2.94) [0.00,3.50]
Total: 4930 W: 791 L: 814 D: 3325
sprt @ 60+0.6 th 1 Shuffle limit : decrease by 12 if pieces count > 14 && verify (tte.depth > 6, abs(ttValue) < 600)
19-04-11 MJZ Shuffle-Tune3 diff
LLR: -1.93 (-2.94,2.94) [0.00,3.50]
Total: 85847 W: 14604 L: 14476 D: 56767
sprt @ 60+0.6 th 1 Shuffle limit : decrease by 12 if pieces count > 12.
19-04-12 xot lever1 diff
LLR: -2.96 (-2.94,2.94) [0.00,3.50]
Total: 22531 W: 3738 L: 3851 D: 14942
sprt @ 60+0.6 th 1 LTC: add !phalanx test.
19-04-11 31m cpasser_supporting^^ diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 78183 W: 13350 L: 13304 D: 51529
sprt @ 60+0.6 th 1 Speculative LTC for 70K yellow. Low throughput (500).