Stockfish Testing Queue

Finished - 29280 tests

16-04-07 n_p Combo4 diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 19270 W: 3664 L: 3440 D: 12166
sprt @ 10+0.1 th 1 Combo patch of two tuning patches which finished yellow at LTC.
16-04-07 Mys MO2 diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 22509 W: 4196 L: 3974 D: 14339
sprt @ 10+0.1 th 1 I like VoyagerOne's idea of using piecevalue in move ordering, so I think it will help in evasions as we can make heavy pieces more important this way.
16-04-07 Voy CaptureOrder diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18098 W: 3250 L: 3298 D: 11550
sprt @ 10+0.1 th 1 Take 2
16-04-05 pb0 distinct_captureorder diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 20535 W: 3339 L: 3381 D: 13815
sprt @ 5+0.05 th 5 Lazy SMP, move order: pick best captures among equal scores in different (alternated) order
16-04-06 Voy CaptureOrder diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18958 W: 3363 L: 3408 D: 12187
sprt @ 10+0.1 th 1 Use moved piece as part of capture scoring formula.
16-04-06 Mys mo diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 26613 W: 4850 L: 4861 D: 16902
sprt @ 10+0.1 th 1 Evasion ordering tweak take 2
16-04-06 Voy mcp2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10484 W: 1836 L: 1917 D: 6731
sprt @ 10+0.1 th 1 Sorry forgot to flip the sign as well...
16-04-06 sg fmh_update diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 17906 W: 3222 L: 3271 D: 11413
sprt @ 10+0.1 th 1 Update fmh only if previous move was not a capture
16-04-06 Voy mcp2 diff
LLR: -1.42 (-2.94,2.94) [0.00,5.00]
Total: 1148 W: 175 L: 230 D: 743
sprt @ 10+0.1 th 1 Do the opposite...
16-04-06 Voy mcp diff
LLR: -2.05 (-2.94,2.94) [0.00,5.00]
Total: 3528 W: 597 L: 669 D: 2262
sprt @ 10+0.1 th 1 Reduce move count by 1 for move count pruning...if there were no captures played yet.
16-04-02 jos bishop_pair diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 135629 W: 18089 L: 17878 D: 99662
sprt @ 60+0.6 th 1 LTC: Is the pair of bishops yet more valuable?
16-04-06 Hai passed diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 21993 W: 4197 L: 4078 D: 13718
sprt @ 10+0.1 th 1 Small passed pawn simplification.
16-04-06 Voy QE diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8001 W: 1411 L: 1502 D: 5088
sprt @ 10+0.1 th 1 Queen Evasion. Evade attacking pawns.
16-04-06 jhe nonpvtt diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 14836 W: 2624 L: 2686 D: 9526
sprt @ 10+0.1 th 1 Do not use TT for non-pv nodes when in check.
16-04-05 Voy sneakyKnight3c diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 14430 W: 2589 L: 2653 D: 9188
sprt @ 10+0.1 th 1 Just realized I should be using pos.see() and not pos.sign_see().
16-04-03 Roc UndefendedKR diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 36805 W: 5100 L: 4857 D: 26848
sprt @ 60+0.6 th 1 This test was yellow at STC and relates to King safety. See if it scales at LTC. Low throughput and will cancel after 10000 if not positive.
16-04-05 Voy sneakyKnight3b diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10147 W: 1781 L: 1863 D: 6503
sprt @ 10+0.1 th 1 Fixed faulty logic...
16-04-05 Elb attack_units diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 30741 W: 5612 L: 5663 D: 19466
sprt @ 10+0.1 th 1 Tweak attack units values (Take 2)
16-04-05 Voy sneakyKnight3 diff
LLR: -2.09 (-2.94,2.94) [0.00,5.00]
Total: 5294 W: 928 L: 994 D: 3372
sprt @ 10+0.1 th 1 I believe this version will be much better. (I hope)
16-04-05 Mys RI diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 23407 W: 4283 L: 4308 D: 14816
sprt @ 10+0.1 th 1 Refreshed take on pawn blocks
16-04-05 Voy sneakyKnight2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 11889 W: 2083 L: 2158 D: 7648
sprt @ 10+0.1 th 1 Take 2
16-04-04 Elb attack_units diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 55217 W: 10021 L: 9989 D: 35207
sprt @ 10+0.1 th 1 Tweak attack units values
16-04-04 Mys npp diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 44956 W: 8195 L: 8128 D: 28633
sprt @ 10+0.1 th 1 Take 3 skewed penalty and no push because of enemy pawn.
16-04-05 pb0 captures_expectance diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18275 W: 3348 L: 3395 D: 11532
sprt @ 10+0.1 th 1 take3, relegate first good capture to last good capture if SEE == 0
16-04-05 pb0 captures_expectance diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16832 W: 3005 L: 3059 D: 10768
sprt @ 10+0.1 th 1 take2, move order: swap moves if 2nd ranked capture expectation is better than SEE score of first
16-04-05 Voy sneakyKnight diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 22521 W: 4098 L: 4127 D: 14296
sprt @ 10+0.1 th 1 A Quick & Dirty attempt-1 at queen evasion from sneaky knights. I believe queen evasions should be new move stage as this will significantly improve move ordering...
16-04-05 Hai singular_reduce diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 17869 W: 3257 L: 3306 D: 11306
sprt @ 10+0.1 th 1 When a move is singular extended increase reduction of remaining moves.
16-04-04 IIv new_tmm diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 2538 W: 392 L: 554 D: 1592
sprt @ 15+0 th 1 One more sudden death check, and then I continue with thorough local work.
16-04-04 Voy smp diff
LLR: -1.11 (-2.94,2.94) [0.00,5.00]
Total: 2609 W: 300 L: 336 D: 1973
sprt @ 30+0.3 th 3 Check to see if having a dedicated cmh/fmh per thread is beneficial for SMP. (Will stop test early if the ELO gain/loss is obvious.)
16-04-04 IIv new_tmm diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 287 W: 14 L: 181 D: 92
sprt @ 40/10 th 1 First try on 40/10 time controls.
16-04-03 IIv new_tmm diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 50717 W: 9068 L: 9311 D: 32338
sprt @ 10+0.1 th 1 This doesn't cover all cases, but I would like to see this simple time management on STC.
16-04-04 IIv new_tmm diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 2031 W: 296 L: 456 D: 1279
sprt @ 15+0 th 1 Before finding better constants, and solving movestogo case, I would like to check sudden death performance.
16-04-04 pb0 captures_expectance diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12708 W: 2268 L: 2339 D: 8101
sprt @ 10+0.1 th 1 move order: Swap moves if 2nd ranked capture returns a better SEE score than the first, take1
16-04-04 Mys npp diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11502 W: 2083 L: 2159 D: 7260
sprt @ 10+0.1 th 1 Take 2 EG penalty instead
16-04-04 Mys sft diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 35066 W: 6387 L: 6362 D: 22317
sprt @ 10+0.1 th 1 Target only non-queen specific endgames for scale factoring
16-04-04 mco lsb32^ diff
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 28548 W: 5308 L: 5060 D: 18180
sprt @ 10+0.1 th 1 32-bit lsb: Use home grown software implementation, as suggested by Joona. Reschedule with STC time control.
16-04-04 Mys npp diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12687 W: 2288 L: 2359 D: 8040
sprt @ 10+0.1 th 1 MG penalty if no available pawn pushes.
16-04-03 sg fmh diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 27644 W: 5060 L: 5122 D: 17462
sprt @ 10+0.1 th 1 try average of cmh and fmh for move order
16-04-04 lbr lsb32^ diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 19190 W: 4071 L: 4160 D: 10959
sprt @ 4+0.04 th 1 32-bit lsb: Use home grown software implementation, as suggested by Joona. Affects 32-bit workers, where it is supposed to be faster. I don't know if there are enough on fishtest to tip the scale, but let's see... Using fast tc to increase the sensitivity of the measure. Here we do not care about scaling. Just a non functional speed-up. @VoyagerOne: Throughput=2500 in 4+0.04 is equivalent to 1000 in 10+0.1, so please don't touch my throughput.
16-04-04 lbr lsb32 diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 9066 W: 2015 L: 1818 D: 5233
sprt @ 4+0.04 th 1 32-bit lsb: Use two 32-bit BSF to recompose 64-bit BSF. Affects 32-bit workers only, where it is supposed to be faster. I don't know if there are enough on fishtest to tip the scale, but let's see... Using fast tc to increase the sensitivity of the measure. Here we do not care about scaling. Just a non functional speed-up. @VoyagerOne: Throughput=2500 in 4+0.04 is equivalent to 1000 in 10+0.1, so please don't touch my throughput.
16-04-04 Voy lmrQS2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7120 W: 1246 L: 1341 D: 4533
sprt @ 10+0.1 th 1 Take 2.... dive into QS if bad history or cut-node.
16-04-03 Voy lmrDynamic diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10465 W: 1864 L: 1945 D: 6656
sprt @ 10+0.1 th 1 Take 2 (fixed bench)
16-04-03 Voy lmrQS diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5046 W: 878 L: 982 D: 3186
sprt @ 10+0.1 th 1 Dive into QS at LMR only at certain conditions.
16-04-03 Voy lmrQS diff
LLR: -0.32 (-2.94,2.94) [0.00,5.00]
Total: 3207 W: 572 L: 572 D: 2063
sprt @ 10+0.1 th 1 Fixed faulty logic
16-04-03 mbo tuned_pp diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 11019 W: 2024 L: 2143 D: 6852
sprt @ 10+0.1 th 1 Test tuned PP and endgame material values. (Corrected bounds)
16-04-03 sg fmh diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5745 W: 1004 L: 1105 D: 3636
sprt @ 10+0.1 th 1 try maximum of cmh and fmh for move order
16-04-03 mbo tuned_pp diff
LLR: 0.02 (-2.94,2.94) [0.00,5.00]
Total: 49 W: 10 L: 9 D: 30
sprt @ 10+0.1 th 1 Test tuned PP and endgame material values.
16-04-02 mbo tune_pp2 diff
24687/25000 iterations
50000/50000 games played
50000 @ 20+0.2 th 1 Tune PP and endgame material.
16-04-01 Roc BackwardSG2 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 40810 W: 5619 L: 5526 D: 29665
sprt @ 60+0.6 th 1 Running a rewrite of SG idea as a simplification against current master (No functional change over his test)
16-03-29 lbr onepawn diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 132604 W: 12828 L: 13105 D: 106671
sprt @ 60+0.6 th 1 LTC: Retire SCALE_FACTOR_ONEPAWN