Stockfish Testing Queue

Finished - 35502 tests

16-03-07 Voy lmrTweak2 diff
LLR: -1.12 (-2.94,2.94) [0.00,5.00]
Total: 1037 W: 161 L: 204 D: 672
sprt @ 10+0.1 th 1 Take 2.
16-03-06 Voy lmrTweak diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 46132 W: 8597 L: 8522 D: 29013
sprt @ 10+0.1 th 1 STC
16-03-06 IIv Rf_delta diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 17487 W: 3177 L: 3274 D: 11036
sprt @ 10+0.1 th 1 Rf+delta, take2.
16-03-06 sni mobility2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 19450 W: 3605 L: 3646 D: 12199
sprt @ 10+0.1 th 1 Take 2, try without the empty square condition
16-03-06 sni mobility2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7104 W: 1256 L: 1351 D: 4497
sprt @ 10+0.1 th 1 Endgame bonus for each square controlled in the opponent side by our pawns but not by the opponent. With bug fixed by Rocky640.
16-03-06 IIv Rf_delta diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 24457 W: 4441 L: 4514 D: 15502
sprt @ 10+0.1 th 1 Delta value seems pretty asymmetric for tuning, but I got some clues. I plan to try (2.20, 18), (2.15, 17), (2.05, 16), but it's possible that I'll change my mind depending on the results.
16-03-06 IIv search_tune diff
38429/40000 iterations
78450/80000 games played
80000 @ 10+0.1 th 1 I didn't plan to work on this any more, but I think that recent simplification in reduction formula deserves this try. Low throughput and corrected code.
16-03-05 Voy mcptb diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 40747 W: 5363 L: 5403 D: 29981
sprt @ 60+0.6 th 1 If approvers don't mind, I would like to try this yellow patch at LTC. Local testing is promising and pruning tweaks can be quite TC sensitive. Low throughput.
16-03-05 jos ENDGAMES1 diff
ELO: -2.02 +-3.1 (95%) LOS: 10.1%
Total: 10000 W: 1009 L: 1067 D: 7924
10000 @ 60+0.6 th 1 Some rework/simplification of endgames and scaling factor. Do a quick measure at LTC and with 8moves book.
16-03-06 fbv RookAlignedPawns diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 16878 W: 3019 L: 3072 D: 10787
sprt @ 10+0.1 th 1 Before aligning with enemy pawns look mobility
16-03-06 sni initiative5 diff
LLR: -1.41 (-2.94,2.94) [0.00,5.00]
Total: 711 W: 94 L: 150 D: 467
sprt @ 10+0.1 th 1 Are blocked pawns of any use in evaluate_initiative() ?
16-03-05 IIv search_tune diff
11510/40000 iterations
23439/80000 games played
80000 @ 10+0.1 th 1 I didn't plan to work on this any more, but I think that recent simplification in reduction formula deserves this try. Low throughput
16-03-05 fbv LastRookMobility diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9676 W: 1767 L: 1851 D: 6058
sprt @ 10+0.1 th 1 Last try. Hope the bench is ok.
16-03-05 Roc LowRankBishop diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5499 W: 962 L: 1064 D: 3473
sprt @ 10+0.1 th 1 Take 2
16-03-05 fbv RookAlignedPawns1 diff
LLR: -0.23 (-2.94,2.94) [0.00,5.00]
Total: 178 W: 30 L: 39 D: 109
sprt @ 10+0.1 th 1 Before aligning with enemy pawns look is not too cramped
16-03-05 sni initiative5 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 47015 W: 8554 L: 8550 D: 29911
sprt @ 10+0.1 th 1 Change the initiative offset (128)
16-03-05 Elb tr_param diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 42090 W: 7973 L: 7982 D: 26135
sprt @ 10+0.1 th 1 Tweak TrappedRook parameter
16-03-05 sni initiative5 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 20077 W: 3674 L: 3762 D: 12641
sprt @ 10+0.1 th 1 Change initiative offset (112)
16-03-05 fbv RookMobility06 diff
LLR: -0.54 (-2.94,2.94) [0.00,5.00]
Total: 530 W: 90 L: 111 D: 329
sprt @ 10+0.1 th 1 A lower penalty and mob == 0
16-03-05 Roc LowRankBishop diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 11572 W: 2095 L: 2171 D: 7306
sprt @ 10+0.1 th 1 Take 1
16-03-05 Roc WeakNearKing diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12872 W: 2374 L: 2444 D: 8054
sprt @ 10+0.1 th 1 Take 3
16-03-05 Elb blocked_lines diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10269 W: 1867 L: 1948 D: 6454
sprt @ 10+0.1 th 1 Reduced penalty and also improve trap detection.
16-03-04 sni simplify_checks diff
LLR: -3.12 (-2.94,2.94) [-3.00,1.00]
Total: 72497 W: 13342 L: 13635 D: 45520
sprt @ 10+0.1 th 1 Simplify safe checks analysis : is the loop slowdown acceptable? Same bench.
16-03-04 Voy mcptb diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 78826 W: 14372 L: 14259 D: 50195
sprt @ 10+0.1 th 1 mcp tweak
16-03-05 IIv new_tmm diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 31145 W: 5636 L: 5847 D: 19662
sprt @ 10+0.1 th 1 I guess a tuned value will not work well, but let's check. Anyway, my second last attempt with time management simplification.
16-03-04 Elb blocked_lines diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 27096 W: 5019 L: 5027 D: 17050
sprt @ 10+0.1 th 1 Take 3: increased penalty and slight speedup
16-03-04 Voy cm-pc3 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 7702 W: 1384 L: 1477 D: 4841
sprt @ 10+0.1 th 1 Last shot at this idea...
16-03-04 Elb blocked_lines diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 60398 W: 11248 L: 11112 D: 38038
sprt @ 10+0.1 th 1 Take 2: now hopefully with correct check for not attacking pieces
16-03-04 Roc WeakNearKing diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 21999 W: 4067 L: 4097 D: 13835
sprt @ 10+0.1 th 1 Take 2.
16-03-04 jos ENDGAMES2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11915 W: 2168 L: 2242 D: 7505
sprt @ 10+0.1 th 1 Test 3 cherry-picked from ENDGAMES1 improvements (hopefully).
16-03-04 fbv RookMobility05 diff
LLR: 0.85 (-2.94,2.94) [0.00,5.00]
Total: 12906 W: 2414 L: 2322 D: 8170
sprt @ 10+0.1 th 1 Caged rook. Optimized code and new (lower) penalty
16-03-04 sni forks5 diff
LLR: -2.48 (-2.94,2.94) [0.00,4.00]
Total: 8683 W: 1542 L: 1644 D: 5497
sprt @ 10+0.1 th 1 Rewrite safe checks analysis as a loop, and introduce fork analysis in the loop. Tested with [0..4] bounds because this is a 13 lines simplification of current master.
16-03-04 sni forks5 diff
LLR: -2.45 (-2.94,2.94) [0.00,4.00]
Total: 7973 W: 1424 L: 1527 D: 5022
sprt @ 10+0.1 th 1 Take 2
16-03-04 Elb blocked_lines diff
LLR: -2.20 (-2.94,2.94) [0.00,5.00]
Total: 5465 W: 967 L: 1037 D: 3461
sprt @ 10+0.1 th 1 Give small penalty when rook isn't attacking anything and it can only move on a line or rank
16-03-04 IIv tmm_tune diff
24719/25000 iterations
50000/50000 games played
50000 @ 10+0.1 th 1 Tuning ratio 2 without unstable non-linear term. Local tests showed ill-behaviour of introducing pow function in ratio2 for time management simplification (corrected bench).
16-03-04 Elb rook_trap diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 28231 W: 5270 L: 5273 D: 17688
sprt @ 10+0.1 th 1 Take 2: add attacked penalty
16-03-04 fbv RookMobility04 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13195 W: 2381 L: 2450 D: 8364
sprt @ 10+0.1 th 1 Saving caged rook. mob=0. Take 2
16-03-04 pb0 history2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 6948 W: 1222 L: 1318 D: 4408
sprt @ 10+0.1 th 1 Introducing a 2nd layer in history for better move ordering for plies near the root, take2. (This passed a local SPRT[0,5] last night but my machines seem to be not very reliable in that sense)
16-03-04 Roc WeakNearKing diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 12530 W: 2293 L: 2365 D: 7872
sprt @ 10+0.1 th 1 Penalize more some weak pawns near our King
16-03-04 fbv RookMobility03 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 33649 W: 6217 L: 6197 D: 21235
sprt @ 10+0.1 th 1 Try 3. Condition change
16-03-03 Roc RookRookMob diff
LLR: -2.98 (-2.94,2.94) [0.00,5.00]
Total: 8721 W: 1561 L: 1650 D: 5510
sprt @ 10+0.1 th 1
16-03-03 sni loose2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 39010 W: 7157 L: 7115 D: 24738
sprt @ 10+0.1 th 1 Bonus for opponent loose pieces. Take 4, bonus = S(20,20)
16-03-03 fbv RookMobility01b diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 33401 W: 4539 L: 4545 D: 24317
sprt @ 60+0.6 th 1 LTC. Based on ElbertoOne's idea. Now adding rank_bb
16-03-03 jos ENDGAMES1 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 36274 W: 6571 L: 6791 D: 22912
sprt @ 10+0.1 th 1 Some rework/simplification of endgames and scaling factor. Test as simplification, though this should be tested without adjudication rules, imho. Just see if it fails or pass.
16-03-03 sni loose2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 36196 W: 6634 L: 6604 D: 22958
sprt @ 10+0.1 th 1 Bonus for opponent loose pieces. Take 2, bonus = S(0,30)
16-03-03 Elb rook_trap diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 12679 W: 2324 L: 2395 D: 7960
sprt @ 10+0.1 th 1 Another rook mobility try. Increased penalty.
16-03-03 Voy PV-HS diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 4651 W: 802 L: 908 D: 2941
sprt @ 10+0.1 th 1 History Split based on PvNode
16-03-03 fbv RookMobility02 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 28316 W: 5261 L: 5264 D: 17791
sprt @ 10+0.1 th 1 Try 02. Only if RookMobility01b fails. Rank limits between 2 and 7
16-03-03 Voy cm-pc2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10204 W: 1798 L: 1880 D: 6526
sprt @ 10+0.1 th 1 Take 2
16-03-03 sni loose2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 4191 W: 729 L: 837 D: 2625
sprt @ 10+0.1 th 1 Bonus for opponent loose pieces. Take 3, bonus = S(0,15)