Stockfish Testing Queue

Pending - 0 tests 0.0 hrs

None

Active - 7 tests

18-05-28 And ks_weak_by_2 diff
LLR: 0.38 (-2.94,2.94) [0.00,5.00]
Total: 865 W: 193 L: 172 D: 500
sprt @ 10+0.1 th 1 More penalty for the weak squares attacked by two.
18-05-28 And ks_attacked_by_2 diff
LLR: 0.27 (-2.94,2.94) [0.00,5.00]
Total: 528 W: 111 L: 97 D: 320
sprt @ 10+0.1 th 1 Penalty for squares attacked by 2.
18-05-28 And ks_king_ring_pawns diff
LLR: -0.17 (-2.94,2.94) [0.00,5.00]
Total: 511 W: 89 L: 94 D: 328
sprt @ 10+0.1 th 1 More safe with pawns in king ring. Adjust baseline to compensate.
18-05-28 And ks_king_ring_pawns2 diff
LLR: -0.26 (-2.94,2.94) [0.00,5.00]
Total: 523 W: 100 L: 109 D: 314
sprt @ 10+0.1 th 1 Raise values from version 1.
18-05-28 And ks_weak_by_2A diff
LLR: -0.29 (-2.94,2.94) [0.00,5.00]
Total: 538 W: 99 L: 109 D: 330
sprt @ 10+0.1 th 1 Reduce normal weak weight, add to the weak_by_2 weight. bench Ÿ5104075
18-05-28 And ks_king_area_scale9 diff
LLR: -0.37 (-2.94,2.94) [0.00,5.00]
Total: 681 W: 125 L: 138 D: 418
sprt @ 10+0.1 th 1 Scale king area for attack counts. "
18-05-28 And ks_king_area_scale11 diff
LLR: -0.47 (-2.94,2.94) [0.00,5.00]
Total: 461 W: 81 L: 99 D: 281
sprt @ 10+0.1 th 1 Same as ks_king_area_scale9, but use 11.

Finished - 370 tests

18-05-28 And ks_pawn_attackcounts diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 12165 W: 2447 L: 2563 D: 7155
sprt @ 10+0.1 th 1 Add the pawn attacks to kingAttackCounts as well.
18-05-25 And smp_aspiration diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 20767 W: 3710 L: 3747 D: 13310
sprt @ 5+0.05 th 8 Vary window size per thread. Non functional on 1 thread.
18-05-25 And overloaded_pawn_by2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10689 W: 2088 L: 2167 D: 6434
sprt @ 10+0.1 th 1 Overloaded uses attacked and defended just once ... let us attack more than once if we are using a pawn to do so.
18-05-25 And overloaded_pawn_attacks diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10477 W: 2066 L: 2146 D: 6265
sprt @ 10+0.1 th 1 Don't allow pawns to define overloaded pieces.
18-05-25 And mcp_statscore_A diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 35257 W: 7083 L: 7049 D: 21125
sprt @ 10+0.1 th 1 Don't start MCP until we hit bad quiets.
18-05-25 And prob_pick_tt diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13176 W: 2582 L: 2650 D: 7944
sprt @ 10+0.1 th 1 Let an okay probcut move become the ttMove.
18-05-25 And lmr_depth_never_imp diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 12238 W: 2420 L: 2535 D: 7283
sprt @ 10+0.1 th 1 Always use improving=false for lmrDepth.
18-05-25 And mcp_statscore_B diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 1818 W: 314 L: 435 D: 1069
sprt @ 10+0.1 th 1 Same as _A, but use statscore < 0. "
18-05-16 And tt_less_age diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 39794 W: 7994 L: 8007 D: 23793
sprt @ 10+0.1 th 1 Less weight on tt age for replacment.
18-05-07 And skip_razoring_rq diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10621 W: 2110 L: 2189 D: 6322
sprt @ 10+0.1 th 1 Extend sg's patch to include rook captures.
18-05-07 And see_margins_imp diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 50850 W: 10281 L: 10253 D: 30316
sprt @ 10+0.1 th 1 See pruning margins by improving flag.
18-05-05 And prob_mcp_valueB diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 34633 W: 7022 L: 7054 D: 20557
sprt @ 10+0.1 th 1 Double the original patch.
18-05-05 And prob_mcp_valueA diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 63000 W: 12934 L: 12859 D: 37207
sprt @ 10+0.1 th 1 Don't require improving, tweak from 90k yellow.
18-05-05 And stat_scoring_again diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7611 W: 1509 L: 1602 D: 4500
sprt @ 10+0.1 th 1 Retry this 100k yellow [0,5] from before the recent stat score changes.
18-05-03 And prob_see_r diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 34553 W: 6931 L: 6900 D: 20722
sprt @ 10+0.1 th 1 even when eval > rBeta, don't consider bad captures.
18-05-03 And prob_see_no_prelim diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 11094 W: 2187 L: 2369 D: 6538
sprt @ 10+0.1 th 1 See if this can replace preliminary.
18-05-02 And pawn_threat_safe_def diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5872 W: 1154 L: 1256 D: 3462
sprt @ 10+0.1 th 1 Redine what is considered an unsafe square for pawn threats.
18-05-01 And bishop_ranks diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 17668 W: 3572 L: 3619 D: 10477
sprt @ 10+0.1 th 1 For white ranks2-5, for black ranks6-3.
18-05-01 And bishop_center_ranks diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 15869 W: 3166 L: 3221 D: 9482
sprt @ 10+0.1 th 1 Only for blocked pawns Rank3-6.
18-04-28 And prob_mcp_value diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 88698 W: 18082 L: 17913 D: 52703
sprt @ 10+0.1 th 1 Try more moves when improving and already beating beta.
18-04-28 And counter_move_imp2 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 67193 W: 13527 L: 13439 D: 40227
sprt @ 10+0.1 th 1 Opposite. Reduce depth threshold if in an improving node.
18-04-28 And prob_mcp_value2 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 55589 W: 11276 L: 11230 D: 33083
sprt @ 10+0.1 th 1 This time using rBeta instead of beta.
18-04-28 And razoring_tt_noisy diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 28043 W: 5691 L: 5689 D: 16663
sprt @ 10+0.1 th 1 If best was a noisy move and qs failed, return.
18-04-28 And tt_cap_pv_exact diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 20494 W: 4085 L: 4170 D: 12239
sprt @ 10+0.1 th 1 Do ttCapture first, else pvExact.
18-04-28 And prob_cut_mcp_legality diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16094 W: 3236 L: 3290 D: 9568
sprt @ 10+0.1 th 1 Kind of strange, count the illegal moves torwards the prob cut MCP counter. Though is there may be a relationship between positions where the "ideal" captures are illegal, and those which should not be prob cut. Tested on [0, 5] by choice.
18-04-28 And razoring_like_futility diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 13988 W: 2794 L: 2982 D: 8212
sprt @ 10+0.1 th 1 Testing as a simplification as requested. Had an issue submitting, so I had to make a new branch ...
18-04-28 And counter_move_imp diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 9353 W: 1867 L: 1952 D: 5534
sprt @ 10+0.1 th 1 extend counter move pruning to depth 4 when not improving.
18-04-28 And futility_pruning_cutnod diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 9050 W: 1794 L: 1921 D: 5335
sprt @ 10+0.1 th 1 Raise depth bound for cutnodes.
18-04-28 And check_extend_imp diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5053 W: 1008 L: 1114 D: 2931
sprt @ 10+0.1 th 1 Only check extend in improving nodes.
18-04-08 And probcut_no_tt diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 40277 W: 8224 L: 8165 D: 23888
sprt @ 10+0.1 th 1 Don't start with the ttMove in prob cut
18-04-07 And probcut_expectation diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18238 W: 3693 L: 3737 D: 10808
sprt @ 10+0.1 th 1 Skip probcut if ttentry indicates it may fail
18-04-06 And resolutions3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11657 W: 2378 L: 2452 D: 6827
sprt @ 10+0.1 th 1 Fixed bench typo
18-04-06 And resolutions diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7914 W: 1533 L: 1625 D: 4756
sprt @ 10+0.1 th 1 More LMR when resolving root windows
18-04-06 And resolutions2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4287 W: 841 L: 950 D: 2496
sprt @ 10+0.1 th 1 More LMR on fail high, less on fail low.
18-04-04 And old_voyager diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 58926 W: 12018 L: 11873 D: 35035
sprt @ 10+0.1 th 1 Try this old voyager test which allows IID even when we have a ttMove, if said ttMove is from a low enough draft.
18-04-04 And singular_into_check diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13187 W: 2589 L: 2657 D: 7941
sprt @ 10+0.1 th 1 Same order of events, but allow a failed singular to be check extended.
18-04-04 And reorder_extensions diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6056 W: 1159 L: 1259 D: 3638
sprt @ 10+0.1 th 1 Do check extend, then do singular extension. Bench changes because currently a singular can fail, and then not be check extended.
18-04-02 And iid_beats_beta diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 21242 W: 4300 L: 4330 D: 12612
sprt @ 10+0.1 th 1 Only accept the IID results if the reduced search is >= beta. (Take 1)
18-04-02 And lmr_singular diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 27499 W: 5599 L: 5600 D: 16300
sprt @ 10+0.1 th 1 More LMR if we did a singular extension, so long as we are in a nonPvNode or the ttMove held as best
18-04-02 And lmr_singular2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7252 W: 1429 L: 1524 D: 4299
sprt @ 10+0.1 th 1 More LMR if we did a singular extension, so long as we are in a PvNode and the ttMove held as best
18-04-02 And lmr_singular3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 6215 W: 1188 L: 1288 D: 3739
sprt @ 10+0.1 th 1 More LMR if we did a singular extension, so long as we are not using PvExacat
18-03-31 And null_improving diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 24532 W: 4939 L: 5009 D: 14584
sprt @ 10+0.1 th 1 Slightly raise R in NMP in improving node, slightly lower R in NMP in non-improving node.
18-03-31 And razor_old_improving diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 47802 W: 9791 L: 9697 D: 28314
sprt @ 10+0.1 th 1 Treat non-improving nodes as depth 1 nodes in razoring. Use the old layout...
18-03-31 And revert_razoring diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8360 W: 1649 L: 1739 D: 4972
sprt @ 10+0.1 th 1 Revert all razoring changes since Jan1 2018. Should fail, but please humor me.
18-03-31 And razor_faltering diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4231 W: 812 L: 921 D: 2498
sprt @ 10+0.1 th 1 Treat nodes where we are falling behind in the same way as depth one nodes in regards to razoring
18-03-23 And pvExactDepthA diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10578 W: 2109 L: 2189 D: 6280
sprt @ 10+0.1 th 1 Jerry's patch looks pretty strong. My patch which avoided PvExact when we got the tt from IID went red. So many if we move Jerry's margin to include the IID entries, we get a green
18-03-22 And lmrCapturesB diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 29163 W: 5993 L: 5985 D: 17185
sprt @ 10+0.1 th 1 For captures in LMR, RAISE R when we are looking at attacks to DIFFERENT squares as the ttCapture
18-03-22 And ttCaptureDidIID diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 35037 W: 7121 L: 7087 D: 20829
sprt @ 10+0.1 th 1 Don't do ttCapture if the tt entry came from IID
18-03-22 And lmrCapturesA diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 11222 W: 2280 L: 2356 D: 6586
sprt @ 10+0.1 th 1 For captures in LMR, lower R when we are looking at different pieces which attack the same square as the ttCapture
18-03-22 And pvExactDidIID diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 19134 W: 3839 L: 3879 D: 11416
sprt @ 10+0.1 th 1 Don't do pvExact if the tt entry came from IID