Stockfish Testing Queue

Pending - 0 tests 0.0 hrs

None

Active - 0 tests

Finished - 1030 tests

18-06-24 jos simplifyKP diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 46069 W: 7949 L: 7870 D: 30250
sprt @ 60+0.6 th 1 LTC: Simplify KingProtector penalty. Apply penalty only to knights and bishops.
18-06-24 jos simplifyKP diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 20873 W: 4592 L: 4469 D: 11812
sprt @ 10+0.1 th 1 Simplify KingProtector penalty. Apply penalty only to knights and bishops.
18-06-08 jos time_tweak^ diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 60308 W: 12191 L: 12128 D: 35989
sprt @ 10+0.1 th 1 Tweak timeReduction factor 1.20. Take 1.
18-06-08 jos time_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 11271 W: 2220 L: 2339 D: 6712
sprt @ 10+0.1 th 1 Tweak timeReduction factor 1.30. Take 2.
18-05-22 jos scaling_last diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 46540 W: 9166 L: 9412 D: 27962
sprt @ 10+0.1 th 1 Take 2. Move setting of scaling function up yet under the calculation of imbalance eval, and don't return SCALE_FACTOR_NONE.
18-05-22 jos scaling_last diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 8192 W: 1497 L: 1671 D: 5024
sprt @ 10+0.1 th 1 This simplification patch combines several ideas and is a continuation of ceebo's original idea. 1. Make sure the main eval uses consistent data for imbalance eval in any case. Also don't calculate imbalance when material is even. 2. Make sure, e->factor[c] is fully computed before setting a scaling function. The scaling function might return SCALE_FACTOR_NONE with e->factor[c] still set to SCALE_FACTOR_NORMAL. This can happen for instance in KBPKB or KBPPKB endgames. 3. This allows to move the ScaleFactor computation from the main evaluation into the material section where it belongs, imho.
18-05-14 jos mcp2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 1701 W: 295 L: 416 D: 990
sprt @ 10+0.1 th 1 Another try at movecount pruning.
18-05-07 jos reset_scores diff
ELO: 0.43 +-2.2 (95%) LOS: 65.1%
Total: 40000 W: 8046 L: 7997 D: 23957
40000 @ 10+0.1 th 1 Measure the effect of resetting all root move scores at each new iteration. See https://github.com/official-stockfish/Stockfish/pull/1579
18-04-26 jos fullDepthSearch diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 37821 W: 7751 L: 7703 D: 22367
sprt @ 10+0.1 th 1 STC: Do a full depth search if we don't LMR a move.
18-04-06 jos ct_experiment diff
ELO: 171.42 +-3.0 (95%) LOS: 100.0%
Total: 30000 W: 15546 L: 1839 D: 12615
30000 @ 10+0.1 th 1 How is this one doing against a weaker opponent (SF7)? Baseline is fisherman's latest test http://tests.stockfishchess.org/tests/view/5ac4a01f0ebc590305f0f425
18-03-31 jos ct_experiment diff
ELO: -1.68 +-2.1 (95%) LOS: 5.8%
Total: 40000 W: 7464 L: 7657 D: 24879
40000 @ 10+0.1 th 1 A first measurement of a contempt experiment. Can we achieve the same by shifting the draw score instead of the evaluation? (Rescheduled, modifying doesn't put it back into queue ...)
18-03-30 jos ct_experiment diff
ELO: -2.01 +-2.8 (95%) LOS: 7.8%
Total: 22847 W: 4273 L: 4405 D: 14169
40000 @ 10+0.1 th 1 A first measurement of a contempt experiment. Can we achieve the same by shifting the draw score instead of the evaluation?
18-03-26 jos qsft diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 25816 W: 5244 L: 5253 D: 15319
sprt @ 10+0.1 th 1 Take 2.
18-03-26 jos qsft diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 45327 W: 9305 L: 9222 D: 26800
sprt @ 10+0.1 th 1 After the recent changes retest this old idea. Take 1.
18-03-21 jos init1 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 7215 W: 1420 L: 1554 D: 4241
sprt @ 10+0.1 th 1 Take 3.
18-03-20 jos init1 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 7052 W: 1346 L: 1480 D: 4226
sprt @ 10+0.1 th 1 Tuned values (SPSA), take 2.
18-03-18 jos prefetch_probcut diff
ELO: -0.18 +-2.2 (95%) LOS: 43.9%
Total: 27586 W: 4125 L: 4139 D: 19322
30000 @ 60+0.6 th 1 Let's see at LTC with bigger hash if there is any benefit in having the prefetch in ProbCut.
18-03-18 jos prefetch_probcut diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 19086 W: 3881 L: 3921 D: 11284
sprt @ 10+0.1 th 1 Add a prefetch in ProbCut. Not sure this has been tried before.
18-03-16 jos init1 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 19642 W: 4040 L: 4127 D: 11475
sprt @ 10+0.1 th 1 Try some early tuning values for complexity factors.
18-03-14 jos dynamic_ct2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7886 W: 1606 L: 1698 D: 4582
sprt @ 10+0.1 th 1 Change draw scores again. Last try before trying to tune. Test with Contempt=16.
18-03-14 jos dynamic_ct diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18179 W: 3807 L: 3850 D: 10522
sprt @ 10+0.1 th 1 Take 2 with tweaked formula, to always keep some amount of contempt, even when winning/losing. Test with Contempt=20.
18-03-14 jos dynamic_ct diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8888 W: 1970 L: 2058 D: 4860
sprt @ 10+0.1 th 1 Decreasing dynamic contempt based on previous best score. (Even in MultiPV mode all PV lines are now being searched with the same contempt.) Take 1. Test with Contempt=40
18-03-13 jos dynamic_ct diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 20993 W: 4375 L: 4405 D: 12213
sprt @ 10+0.1 th 1 Decreasing dynamic contempt based on previous best score. (Even in MultiPV mode all PV lines are now being searched with the same contempt.) Take 1. Test with Contempt=20
18-02-27 jos multipv_old2 diff
LLR: -2.94 (-2.94,2.94) [-3.00,1.00]
Total: 2420 W: 958 L: 1195 D: 267
sprt @ 10+0.1 th 1 Test the modified multiPV search with Skill Level 10 for no regression. (It looks like 'Skill Level' was not set in the 1st test, so now try without quotation marks..)
18-02-22 jos multipv_old2 diff
LLR: -3.77 (-2.94,2.94) [-3.00,1.00]
Total: 113103 W: 24499 L: 24931 D: 63673
sprt @ 10+0.1 th 1 Test the modified multiPV search with Skill Level 10 for no regression. (Not sure the quotation marks around 'Skill Level' are needed; in local testing they are.)
18-02-25 jos distribute_rm1 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 1886 W: 336 L: 457 D: 1093
sprt @ 5+0.05 th 7 SMP patch. Odd/even distribution of root moves across the helper threads.
18-02-23 jos mcp1 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 15192 W: 3386 L: 3442 D: 8364
sprt @ 10+0.1 th 1 Take 3. Be only more cautious after previous first or second move.
18-02-23 jos mcp1 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 33712 W: 7410 L: 7375 D: 18927
sprt @ 10+0.1 th 1 Movecount pruning tweak. Less pruning after prior early moves.
18-02-23 jos mcp1 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 1918 W: 351 L: 472 D: 1095
sprt @ 10+0.1 th 1 Take 2. More drastic.
18-02-18 jos remove_dyn_ct diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 89968 W: 19357 L: 19697 D: 50914
sprt @ 10+0.1 th 1 Remove dynamic contempt and explicitly reset 'bestValue' at the start of the MultiPV loop. Current dynamic contempt is buggy in case of a multiPV search because 'bestValue' refers to the previous PV line. (Local simplification test sprt(-4, 0) is looking good so far.)
18-02-06 jos multipv_old2 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 149486 W: 33049 L: 33181 D: 83256
sprt @ 10+0.1 th 1 Finally, test for no regression in SinglePV mode. STC only.
18-02-06 jos multipv_old2 diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 3147 W: 1182 L: 1004 D: 961
sprt @ 60+0.6 th 1 LTC: Retest the bugfixed version.
18-02-06 jos multipv_old2 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 1741 W: 726 L: 553 D: 462
sprt @ 10+0.1 th 1 Retest the bugfixed version.
18-02-01 jos multipv_old2 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 2260 W: 848 L: 679 D: 733
sprt @ 60+0.6 th 1 LTC: Let's check the 'old' multipv search as suggested by Stefano.
18-02-01 jos multipv_old2 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 3113 W: 1248 L: 1064 D: 801
sprt @ 10+0.1 th 1 Let's check the 'old' multipv search as suggested by Stefano.
18-01-25 jos outpost_xp diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 10576 W: 2333 L: 2456 D: 5787
sprt @ 10+0.1 th 1 Values changed significantly, so retest. Take 2.
18-01-21 jos cutNode3 diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 6474 W: 1061 L: 1158 D: 4255
sprt @ 10+0.1 th 1 If the reduced depth search fails high, set cutNode to true. Final try.
18-01-21 jos cutNode3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2210 W: 341 L: 457 D: 1412
sprt @ 10+0.1 th 1 One last try, please. We only do a full depth search for the first move, hoping for a cutoff, or if the reduced depth search failed high. Is it safe to assume for both cases that the child node is an ALL node?
18-01-20 jos cutNode3 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 5415 W: 917 L: 1055 D: 3443
sprt @ 10+0.1 th 1 Do the reduced depth search with !cutNode just like the full depth search. This will most likely fail because of doing less reductions now. Last take on this series.
18-01-20 jos cutNode3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 4800 W: 807 L: 912 D: 3081
sprt @ 10+0.1 th 1 If we didn't fail-high at an expected after the first 3 moves, change node type. Take 3.
18-01-20 jos outpost_xp diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 7170 W: 1290 L: 1423 D: 4457
sprt @ 10+0.1 th 1 Yet another tuning experiment.
18-01-19 jos cutNode3 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 16769 W: 2969 L: 3068 D: 10732
sprt @ 10+0.1 th 1 Set cutNode to false for the null-move search. Take 2. (Test as tuning patch.)
18-01-19 jos cutNode3 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 1464 W: 204 L: 323 D: 937
sprt @ 10+0.1 th 1 Change node type if the null-move fails.
18-01-05 jos weak_backrank diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12686 W: 2244 L: 2315 D: 8127
sprt @ 10+0.1 th 1 Take 2. Double penalty.
18-01-04 jos weak_backrank diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 24746 W: 4444 L: 4464 D: 15838
sprt @ 10+0.1 th 1 My try on a penalty for a weak backrank.
17-12-25 jos imbal_fine_epd diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 15324 W: 2717 L: 2821 D: 9786
sprt @ 10+0.1 th 1 Fine tuned values after 225k games.
17-12-13 jos isolated diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 11392 W: 2077 L: 2195 D: 7120
sprt @ 10+0.1 th 1 CLOP tuning try with a big set of positions as used by Texel tuning method instead of an opening book.
17-12-12 jos remove_doubled diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 24098 W: 4276 L: 4474 D: 15348
sprt @ 10+0.1 th 1 Is the current handling of doubled pawns still useful?
17-12-10 jos ext_iter6 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13755 W: 2424 L: 2491 D: 8840
sprt @ 10+0.1 th 1 Always extend checks close to the root. Take 2.
17-12-10 jos ext_iter6 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 19794 W: 3558 L: 3599 D: 12637
sprt @ 10+0.1 th 1 Extend checks during the first iterations even with bad SEE. Take 1.