Stockfish Testing Queue

Pending - 0 tests 0.0 hrs

None

Active - 0 tests

Finished - 192 tests

09-11-17 An razoring_into_1ply_less diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 23895 W: 4227 L: 4251 D: 15417
sprt @ 10+0.1 th 1 Let razoring just drop into a one ply reduced search if we have a quiet ttmove
09-11-17 An razoring_non_pawn_mater diff
LLR: -1.83 (-2.94,2.94) [0.00,5.00]
Total: 12386 W: 2193 L: 2218 D: 7975
sprt @ 10+0.1 th 1 Don't do razoring without non pawn material
09-11-17 An razoring_into_1ply_sear diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 11367 W: 2098 L: 2175 D: 7094
sprt @ 10+0.1 th 1 Let razoring just drop into a 1ply search if we have a quiet ttmove
09-11-17 An razoring_quiet_ttmoves diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9467 W: 1704 L: 1789 D: 5974
sprt @ 10+0.1 th 1 Don't do razoring when we have a quiet ttmove
09-11-17 An contested_isolation_1 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 22885 W: 4065 L: 4094 D: 14726
sprt @ 10+0.1 th 1 Revisit this after a simplification. Increase penalty for an isolated pawn that is attacked by an enemy pawn.
09-11-17 An contested_isolation_2 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 20797 W: 3755 L: 3792 D: 13250
sprt @ 10+0.1 th 1 Revisit this after a simplification. Decrease penalty for an isolated pawn that is attacked by an enemy pawn.
09-11-17 An stat_extension_1 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 17874 W: 3136 L: 3186 D: 11552
sprt @ 10+0.1 th 1 Not long ago we added case for when statscore was bad and ss-1's statscore was not set. Try adding the case for when statscore is good, and ss-1's statscore is not set.
08-11-17 An quiet_stats_when_noisy diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 17983 W: 3251 L: 3299 D: 11433
sprt @ 10+0.1 th 1 Do stat updates for quiet moves when the best move is quiet, or was noisy and found with MCP set
08-11-17 An capture_stats_mcp diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 13988 W: 2518 L: 2584 D: 8886
sprt @ 10+0.1 th 1 Extra stat bonus/malus when best move is a capture that may be have subject to LMR
08-11-17 An regression_pruning2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 3463 W: 580 L: 691 D: 2192
sprt @ 10+0.1 th 1 Take 2. Depths more similar to IID, more strict margins.
08-11-17 An regression_pruning diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 5076 W: 848 L: 952 D: 3276
sprt @ 10+0.1 th 1 Trying pruning based on poor pathways
06-11-17 An poor_tt_move diff
LLR: -2.54 (-2.94,2.94) [0.00,5.00]
Total: 22215 W: 3947 L: 3961 D: 14307
sprt @ 10+0.1 th 1 Get a better ttMove in IID if the ttm came from qsearch, and was noisy.
03-11-17 An stat_mcp diff
LLR: -3.23 (-2.94,2.94) [0.00,5.00]
Total: 55407 W: 9918 L: 9821 D: 35668
sprt @ 10+0.1 th 1 Extra stat bonus if best move was found while move count pruning was set. This has tested well for quiets before, I expect it to be good for capture stats.
03-11-17 An stat_early diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 11979 W: 2089 L: 2164 D: 7726
sprt @ 10+0.1 th 1 Extra stats if no early pruning. Was a 73K yellow, see if new capture stats makes this green.
01-11-17 An cutNodeMCP diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 9327 W: 1646 L: 1732 D: 5949
sprt @ 10+0.1 th 1 STC
31-10-17 An statsBonusMCP diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 75181 W: 13558 L: 13365 D: 48258
sprt @ 10+0.1 th 1 Stat bonus for quiet moves that were found when move count pruning was set. No extra malus for the other quiets.
31-10-17 An statSkipEarlyBonus diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 34749 W: 6259 L: 6237 D: 22253
sprt @ 10+0.1 th 1 Increase stat bonus to quiet move found when we skipped Early Pruning. Do not increase malus to bad moves.
31-10-17 An statBonusMCP2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 22239 W: 3968 L: 3999 D: 14272
sprt @ 10+0.1 th 1 (Take 2) Only increase malus to non best moves.
30-10-17 An ttstats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 2709 W: 445 L: 559 D: 1705
sprt @ 10+0.1 th 1 For tt stat malus, only apply if the move is actually a ttmove.
29-10-17 An pvFutile diff
LLR: -3.44 (-2.94,2.94) [0.00,5.00]
Total: 54687 W: 9771 L: 9686 D: 35230
sprt @ 10+0.1 th 1 Use different futility margin for PvNodes
27-10-17 An reductionsOnCaptures1 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 38307 W: 6814 L: 6778 D: 24715
sprt @ 10+0.1 th 1 Apply two of the R reductions to noisy moves in LMR.
27-10-17 An reductionsOnCaptures2 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 15917 W: 2842 L: 2900 D: 10175
sprt @ 10+0.1 th 1 (Take 2) Remove that -ONE_PLY that is applied to all noisy moves in LMR
26-10-17 An extendStatsPreLMR2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 31610 W: 5623 L: 5615 D: 20372
sprt @ 10+0.1 th 1 Extend stat R reductions to moves prior to setting (ss-1)->statScore (take 2)
26-10-17 An noStatUpdateFollowingNu diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 9581 W: 1682 L: 1767 D: 6132
sprt @ 10+0.1 th 1 Don't update stats after a NULL move..
26-10-17 An extendStatsPreLMR diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 9885 W: 1737 L: 1820 D: 6328
sprt @ 10+0.1 th 1 Extend stat R reductions to moves prior to setting (ss-1)->statScore (take 1)
26-10-17 An nullNonCut diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 19520 W: 3509 L: 3551 D: 12460
sprt @ 10+0.1 th 1 Don't do the 'escapes capture' R reduction for moves following a NULL move.
26-10-17 An statnull2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 48599 W: 8657 L: 8578 D: 31364
sprt @ 10+0.1 th 1 Extra R increase for stat moves that following a null move
26-10-17 An statnull diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8792 W: 1526 L: 1614 D: 5652
sprt @ 10+0.1 th 1 No R reduction for stat moves that following a null move
19-10-17 An noTThitA diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11247 W: 2010 L: 2087 D: 7150
sprt @ 10+0.1 th 1 Combine both of Voyager's noTThit (Take 1)
19-10-17 An noTThitB diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11145 W: 1933 L: 2011 D: 7201
sprt @ 10+0.1 th 1 Combine both of Voyager's noTThit (Take 2)
18-10-17 An statDoFull diff
LLR: -2.23 (-2.94,2.94) [0.00,5.00]
Total: 30665 W: 5423 L: 5389 D: 19853
sprt @ 10+0.1 th 1 Stat depth increase if we did a full search (either a failed reduction, or best move did not qualify for LMR)
18-10-17 An LMRandLMPtweak diff
LLR: -2.94 (-2.94,2.94) [0.00,4.00]
Total: 15612 W: 2718 L: 2821 D: 10073
sprt @ 10+0.1 th 1 More LMR in PvNodes, less LMP in PvNodes
18-10-17 An statEarly2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 49200 W: 8644 L: 8564 D: 31992
sprt @ 10+0.1 th 1 Double depth increase from statEarly
17-10-17 An statEarly diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 72677 W: 12939 L: 12759 D: 46979
sprt @ 10+0.1 th 1 Stat bonus if we did not do any early move pruning
15-10-17 An captureLMR4B diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13414 W: 2316 L: 2385 D: 8713
sprt @ 10+0.1 th 1 Same as captureLMR4A, but do not do anything different from master for cut nodes
15-10-17 An captureLMR4A diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 8634 W: 1490 L: 1579 D: 5565
sprt @ 10+0.1 th 1 Try this again, remove the new static eval logic.
13-10-17 An skipEarlyPruning5 diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 45877 W: 8152 L: 8084 D: 29641
sprt @ 10+0.1 th 1 Another take since there are no tests queued ATM. Skip Early pruning in quiet LMR for PvNodes which have better evals than search values.
13-10-17 An openRookOnKingFileFixed diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13297 W: 2282 L: 2351 D: 8664
sprt @ 10+0.1 th 1 Fixed version
13-10-17 An openRookOnKingFile diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 33842 W: 6016 L: 5999 D: 21827
sprt @ 10+0.1 th 1 Try something that worked for my engine: Increase bonus for semi/open rook if the rook is on the enemy king's file.
12-10-17 An pawnTropismA diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 45581 W: 8200 L: 8132 D: 29249
sprt @ 10+0.1 th 1 My take on sn's pawn tropism
12-10-17 An pawnTropismB diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5824 W: 988 L: 1088 D: 3748
sprt @ 10+0.1 th 1 Increase king danger if there are pawns attacking the king ring, which are defended by additional pawns.
11-10-17 An lmrCaps1 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18254 W: 3095 L: 3144 D: 12015
sprt @ 10+0.1 th 1 Increase LMR on noisy moves that respond to our opponents first move (only in !improving nodes)
10-10-17 An dontSkipIIDForNMP diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9729 W: 1690 L: 1774 D: 6265
sprt @ 10+0.1 th 1 Don't skip our opponents IID after we make a NULL move
10-10-17 An iidStats2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12876 W: 2221 L: 2292 D: 8363
sprt @ 10+0.1 th 1 Same as iidStats, but only for depths lower than 10, since it worked at VSTC but not STC
10-10-17 An skipEarlyPruningNotIID diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 19002 W: 3303 L: 3349 D: 12350
sprt @ 10+0.1 th 1 skipEarlyPruning2 (green) failed at LTC. I did not note that skipEarlyPruning also means skipIID. Maybe by not skipping IID in my patch, we fix scaling.
06-10-17 An statByReductionsA diff
LLR: -1.95 (-2.94,2.94) [0.00,5.00]
Total: 128000 W: 22812 L: 22359 D: 82829
sprt @ 10+0.1 th 1 Increase bonus for a quiet move beating alpha (PV and NonPV, despite git comment) if it was searched at a reduced depth first. Only do this if the stats would normally be updated.
10-10-17 An iidStats diff
LLR: -1.75 (-2.94,2.94) [0.00,5.00]
Total: 11826 W: 2123 L: 2147 D: 7556
sprt @ 10+0.1 th 1 IID may be a waste of time in cutnodes if our opponents move was really bad. Locally results @ 1s+.01s in commit comment.
09-10-17 An cutnodeStatLMR diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 21521 W: 3709 L: 3745 D: 14067
sprt @ 10+0.1 th 1 R += 1 + ss->statScore > 0, not R+= 2 for cutnodes in LMR
09-10-17 An skipEarlyPruning2A diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 13805 W: 2462 L: 2529 D: 8814
sprt @ 10+0.1 th 1 Same as skipEarlyPruning2A, but apply at near root positions, and remove the random +16's. Expected last attempt at this.
09-10-17 An statresets diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16060 W: 2835 L: 2892 D: 10333
sprt @ 10+0.1 th 1 Reset statscore on each search to zero. Always compare (ss-1) using >= 0 or <= 0. Would like to get rid of the confusing behaviour of statscore.