Stockfish Testing Queue

Pending - 0 tests 0.0 hrs

None

Active - 1 tests

25-02-18 An prelimProbcut2 diff
LLR: 1.20 (-2.94,2.94) [0.00,5.00]
Total: 17174 W: 2942 L: 2823 D: 11409
sprt @ 60+0.6 th 1 LTC. Even if the other passes, this fixes two oddities (depth 5 and return value), and I imagine the maintainers would prefer this one. Priority -1 until (if) prelimProbcut2 passes.

Finished - 283 tests

25-02-18 An prelimProbcut diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 23202 W: 4044 L: 3826 D: 15332
sprt @ 60+0.6 th 1 LTC
25-02-18 An prelimProbcut2 diff
LLR: 4.73 (-2.94,2.94) [0.00,5.00]
Total: 36281 W: 8221 L: 7830 D: 20230
sprt @ 10+0.1 th 1 Same as before except 1) Do not change how we handle ProbCut at the lowest depth (D5) and 2) Do not alter the return bound.
25-02-18 An prelimProbcut diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 9841 W: 2251 L: 2071 D: 5519
sprt @ 10+0.1 th 1 Before a preliminary depth one search when we are looking at possible ProbCuts
10-02-18 An triedProbCut diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 9352 W: 1989 L: 2074 D: 5289
sprt @ 10+0.1 th 1 Skip IID if we thought we could do a probcut
03-02-18 An bestTactical diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2201 W: 439 L: 561 D: 1201
sprt @ 10+0.1 th 1 Instead of ttCapture, do bestWasCapture for lmr, but also require the best so far to be the ttMove, or that a few quiets have been tried.
27-01-18 An failHighExcluded diff
LLR: -0.64 (-2.94,2.94) [0.00,5.00]
Total: 32039 W: 5527 L: 5423 D: 21089
sprt @ 60+0.6 th 1 LTC
27-01-18 An failHighExcluded diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 4836 W: 1114 L: 960 D: 2762
sprt @ 10+0.1 th 1 Skip resolving fail high when the move has been the best move for the last iteration and all other moves in the position fail to beat the same beta margin as the current bestmove
26-01-18 An outpostBonus diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13982 W: 3048 L: 3111 D: 7823
sprt @ 10+0.1 th 1 Only apply defended bonus for outposts if there could be a minor threat to defend against.
25-01-18 An safeOutpost2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12130 W: 2682 L: 2754 D: 6694
sprt @ 10+0.1 th 1 Bonus for outposts which cannot be attacked by knights or bishops ever, also supported by a pawn
25-01-18 An safeOutpost diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11184 W: 2429 L: 2505 D: 6250
sprt @ 10+0.1 th 1 Bonus for outposts which cannot be attacked by knights or bishops ever
25-01-18 An returnNullMoveP diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 20566 W: 4553 L: 4583 D: 11430
sprt @ 10+0.1 th 1 Return beta when pruning via null move pruning (3 of 4 return tests) Prio -1
25-01-18 An returnProbCut diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 26947 W: 5920 L: 5919 D: 15108
sprt @ 10+0.1 th 1 Return beta when pruning via probcut (4 of 4 return tests) Prio -1
25-01-18 An returnFutility diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 15100 W: 3262 L: 3319 D: 8519
sprt @ 10+0.1 th 1 Return beta when pruning via futility pruning (2 of 4 return tests) Prio -1
25-01-18 An returnRazoring diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8974 W: 1960 L: 2047 D: 4967
sprt @ 10+0.1 th 1 Return alpha when pruning via razoring (1 of 4 return tests)
22-01-18 An returnValues diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12831 W: 2246 L: 2317 D: 8268
sprt @ 10+0.1 th 1 Return either alpha or beta when doing early pruning; Expect fail, but the framework has not much going on
22-01-18 An ttGivesCheckAllnode diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 19680 W: 3473 L: 3515 D: 12692
sprt @ 10+0.1 th 1 More LMR when ttmove gives check in all nodes (Last version)
22-01-18 An ttGivesCheckHolds diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11535 W: 2021 L: 2097 D: 7417
sprt @ 10+0.1 th 1 More LMR when ttmove gives check and holds best in pv nodes
22-01-18 An ttGivesCheck diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9960 W: 1761 L: 1844 D: 6355
sprt @ 10+0.1 th 1 More LMR when ttmove gives check
22-01-18 An ttGivesCheckCutnodes diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4698 W: 795 L: 900 D: 3003
sprt @ 10+0.1 th 1 More LMR when ttmove gives check in cutnodes
22-01-18 An nmpRExperiment diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13199 W: 2333 L: 2402 D: 8464
sprt @ 10+0.1 th 1 Instead of (eval-beta)/PawnValue, use best of this eval and last eval (from our perspective)
22-01-18 An tacticalNullMovePruning diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17852 W: 3108 L: 3158 D: 11586
sprt @ 10+0.1 th 1 Extra ONE_PLY for R in Null Move Pruning if opponents last move was a capture.
21-01-18 An allExtended diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 5568 W: 939 L: 1040 D: 3589
sprt @ 10+0.1 th 1 If all moves were extended to depth + ONE_PLY, then we save in the table as depth + ONE_PLY
15-01-18 An singularTTDepth diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9784 W: 1715 L: 1799 D: 6270
sprt @ 10+0.1 th 1 Increase rBeta for singular when ttDepth is larger than depth
15-01-18 An bestWasMCP diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 23358 W: 4112 L: 4139 D: 15107
sprt @ 10+0.1 th 1 Similar form of this was tried a while ago, instead of MCP, we actually use this when skipQuiets is set. IE, procs when we would have skipped a quiet...
13-01-18 An ttbounded_razoring diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 27338 W: 4855 L: 4865 D: 17618
sprt @ 10+0.1 th 1 Allow razoring to fall through to qsearch(alpha, beta) for tactical ttMoves with no expectation of beating the razor bound (Take 1)
13-01-18 An ttbounded_razoring2 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10256 W: 1832 L: 1914 D: 6510
sprt @ 10+0.1 th 1 Allow razoring to fall through to qsearch(alpha, beta) for ttMoves with no expectation of beating the razor bound (Take 2) -- allow quiet and noisy ttmoves
10-01-18 An probcut_refutes_null diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10018 W: 1786 L: 1869 D: 6363
sprt @ 10+0.1 th 1 Fixed Bench -- Try something that worked for me: Allow prob cut to refute a null move.
10-01-18 An ks_mobility2 diff
LLR: -0.11 (-2.94,2.94) [0.00,5.00]
Total: 91 W: 8 L: 12 D: 71
sprt @ 60+0.6 th 1 LTC for "Use mobility difference for king safety. (fixed bug)"
08-01-18 An move_count_tune diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 81797 W: 14793 L: 14671 D: 52333
sprt @ 10+0.1 th 1 Try tuning the move count pruning values using a custom tuner
08-01-18 An rounding diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 24217 W: 4406 L: 4480 D: 15331
sprt @ 10+0.1 th 1 Tweak to lmr formula for improving / not improving
03-01-18 An fh_simple diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 22227 W: 2756 L: 2803 D: 16668
sprt @ 20+0.2 th 7 Simple Fail High (STC/LTC/SMP-STC passed), try SMP LTC
03-01-18 An fh_simple diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 9974 W: 1764 L: 1600 D: 6610
sprt @ 5+0.05 th 7 Simple Fail High (STC/LTC passed), try SMP STC
03-01-18 An probcut_d1B diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 60105 W: 11082 L: 10949 D: 38074
sprt @ 10+0.1 th 1 Verify with a depth 1 search without early pruning before proper probcut search
03-01-18 An probcut_d1 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 17209 W: 3086 L: 3138 D: 10985
sprt @ 10+0.1 th 1 Depth 1 search before full search of a probcut node.
02-01-18 An fh_smp_only diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7804 W: 1237 L: 1329 D: 5238
sprt @ 5+0.05 th 7 Try the fail high idea but only for Helper threads
02-01-18 An fail_highs_new diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 2192 W: 341 L: 457 D: 1394
sprt @ 10+0.1 th 1 (Bench Typo Fixed) Instead of skipping depths on fail high, search extend the current best move. If saving time was not the winner on the failHigh series, maybe this was?
31-12-17 An fh_simple diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 126569 W: 16140 L: 15623 D: 94806
sprt @ 60+0.6 th 1 Original passed version + my passed SMP modifications. Disable for MultiPv and extreme valued positions. LTC 1 Thread
31-12-17 An fh_simple diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 4385 W: 828 L: 686 D: 2871
sprt @ 10+0.1 th 1 Original passed version + my passed SMP modifications. Disable for MultiPv and extreme valued positions. STC 1 Thread
29-12-17 An fh_analysis_final diff
LLR: -1.03 (-2.94,2.94) [0.00,5.00]
Total: 16449 W: 2060 L: 2049 D: 12340
sprt @ 60+0.6 th 1 LTC Half Throughput: Final take with this... Cannot seem to find anything to compete with the 4000 game LTC that passed...
30-12-17 An probcut_allnode diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10391 W: 1838 L: 1919 D: 6634
sprt @ 10+0.1 th 1 Only probcut on allnodes
30-12-17 An probcut_cutnode diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8212 W: 1471 L: 1561 D: 5180
sprt @ 10+0.1 th 1 Only probcut on cutnodes
29-12-17 An fh_fixes diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12390 W: 1578 L: 1656 D: 9156
sprt @ 60+0.6 th 1 Same bench as vdvpassed version. Fixes SMP, Analysis, and MultiPV? Please let vdv approve / reject. I don't want to waste any resources. (LTC 1-Thread)
29-12-17 An fh_fixes diff
LLR: 2.96 (-2.94,2.94) [0.00,5.00]
Total: 3775 W: 730 L: 590 D: 2455
sprt @ 10+0.1 th 1 Same bench as vdvpassed version. Fixes SMP, Analysis, and MultiPV? Please let vdv approve / reject. I don't want to waste any resources. (STC 1-Thread)
29-12-17 An fh_fixes diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 1209 W: 495 L: 332 D: 382
sprt @ 10+0.1 th 1 Same bench as vdvpassed version. Fixes SMP, Analysis, and MultiPV? Please let vdv approve / reject. I don't want to waste any resources. (STC 1-Thread MultiPV=3)
28-12-17 An fh_analysis diff
LLR: 0.31 (-2.94,2.94) [0.00,5.00]
Total: 34061 W: 4377 L: 4256 D: 25428
sprt @ 60+0.6 th 1 Try to make the failhigh patches work with MultiPv and Mate finding. See github comments about Mate finding success. (LTC 1Thread) Prio -1 till STC passes. How can I test MultiPV?
29-12-17 An fh_analysis_final diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 4100 W: 765 L: 625 D: 2710
sprt @ 10+0.1 th 1 Final take with this... Cannot seem to find anything to compete with the 4000 game LTC that passed...
29-12-17 An fh_analysis_var1 diff
LLR: -0.88 (-2.94,2.94) [0.00,5.00]
Total: 17797 W: 2283 L: 2261 D: 13253
sprt @ 60+0.6 th 1 (LTC 1-Thread) for removing lastBestMove condition variant
29-12-17 An fh_analysis_var1 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 1584 W: 622 L: 457 D: 505
sprt @ 10+0.1 th 1 MultiPv for removing lastBestMove condition variant. Should finish quickly, so very little resource usage. Just want to verify I'm not breaking anything.
29-12-17 An fh_analysis_var1 diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 3342 W: 687 L: 547 D: 2108
sprt @ 10+0.1 th 1 Variant of fh_analysis, suggested by IIvec. Remove the lastBestMove condition. See how this fares (STC)
28-12-17 An fh_analysis diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 1708 W: 665 L: 499 D: 544
sprt @ 10+0.1 th 1 Quick test with MultiPV again. Last test had many times losses on one machine.