FauziAkram and Joost VandeVondele
fe7b9b14d2
Implement smoother reduction in time management
...
Implement smoother time reduction in time management by replacing a conditional
assignment with a continuous sigmoid-based function. The updated logic employs
a sigmoid-like function for a more gradual adjustment.
Passed STC:
LLR: 2.93 (-2.94,2.94) <0.00,2.00>
Total: 64448 W: 16838 L: 16492 D: 31118
Ptnml(0-2): 145, 7214, 17207, 7466, 192
https://tests.stockfishchess.org/tests/view/6829dc046ec7634154f99fba
Passed LTC:
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 407340 W: 104458 L: 103408 D: 199474
Ptnml(0-2): 196, 42281, 117664, 43335, 194
https://tests.stockfishchess.org/tests/view/6829fe1b6ec7634154f9a036
closes https://github.com/official-stockfish/Stockfish/pull/6091
No functional change
2025-05-25 21:03:16 +02:00
b1b5893a8e
Minor code improvements
...
- Remove / add empty lines
- fix the `ttcapture` comment
- remove the `bonus` variable for `ttMoveHistory`
- remove unnecessary parentheses / brackets
- refactor the movepick good quiet stage
- rename `endMoves` to `endCur`, as the previous name suggests that it points to the end of all generated moves, which it does not.
closes https://github.com/official-stockfish/Stockfish/pull/6089
No functional change.
Co-Authored-By: xu-shawn <50402888+xu-shawn@users.noreply.github.com >
2025-05-25 20:59:27 +02:00
pb00067 and Joost VandeVondele
f58d923fe0
Simplify & improve stalemate detection
...
Change is functional because now we verify for stalemate also on captures and when not giving check.
Green STC test on stalemate-book
https://tests.stockfishchess.org/tests/view/682d878f6ec7634154f9ad2f
Elo: 2.29 ± 1.3 (95%) LOS: 100.0%
Total: 10000 W: 4637 L: 4571 D: 792
Ptnml(0-2): 2, 132, 4664, 202, 0
nElo: 12.42 ± 6.8 (95%) PairsRatio: 1.51
Green LTC test on stalemate-book
https://tests.stockfishchess.org/tests/view/682daa2d6ec7634154f9ad67
Elo: 0.80 ± 0.8 (95%) LOS: 96.9%
Total: 10000 W: 4727 L: 4704 D: 569
Ptnml(0-2): 0, 64, 4849, 87, 0
nElo: 6.51 ± 6.8 (95%) PairsRatio: 1.36
Passed non-regression test @ LTC
https://tests.stockfishchess.org/tests/view/682dd10d6ec7634154f9adb3
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 148512 W: 38135 L: 38046 D: 72331
Ptnml(0-2): 55, 15759, 42558, 15810, 74
N.B.: The unique concern I have, is that due changes in future a negative SEE
capture see might be returned in GOOD_CAPTURE stage. In this case the assert in
can_move_king_or_pawn() will trigger since we must guarantee that all moves
(also quiets) are generated in movepicker when calling can_move_king_or_pawn().
closes https://github.com/official-stockfish/Stockfish/pull/6088
bench: 2178135
2025-05-25 20:55:49 +02:00
Daniel Monroe and Joost VandeVondele
4f021cab3b
Simplify allNode term in prior countermove
...
Passed simplification STC
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 53632 W: 14008 L: 13805 D: 25819
Ptnml(0-2): 136, 6253, 13869, 6388, 170
https://tests.stockfishchess.org/tests/view/6828f2b26ec7634154f99b5e
Passed simplification LTC
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 82482 W: 21202 L: 21045 D: 40235
Ptnml(0-2): 37, 8986, 23052, 9115, 51
https://tests.stockfishchess.org/tests/view/6829010a6ec7634154f99db3
closes https://github.com/official-stockfish/Stockfish/pull/6068
Bench: 2302782
2025-05-23 08:51:54 +02:00
Nonlinear2 and Joost VandeVondele
54fb42ddf8
clean up code
...
**Non functional changes:**
in search.cpp:
- an unnecessary pair of parenthesis in the IIR condition has been removed.
- refactored the stalemate trap detection code
in movepick.cpp:
- use the variables `from`, `to`, `piece`, `pieceType` and `capturedPiece` instead of calling the same functions multiple times in `MovePicker::score()`.
- rename `MovePicker::other_piece_types_mobile()`.
**Functional changes:**
- make sure the processed move is always legal in `MovePicker::other_piece_types_mobile()`.
passed non regression STC:
https://tests.stockfishchess.org/tests/view/6829da686ec7634154f99faf
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 95680 W: 24962 L: 24820 D: 45898
Ptnml(0-2): 221, 9622, 28025, 9738, 234
Passed non regression LTC:
https://tests.stockfishchess.org/tests/view/682a102c6ec7634154f9a086
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 117666 W: 30065 L: 29957 D: 57644
Ptnml(0-2): 45, 10173, 38291, 10277, 47
Run of 10k games on the stalemate opening book:
https://tests.stockfishchess.org/tests/view/682b114e6ec7634154f9aa2d
Elo: 0.76 ± 0.9 (95%) LOS: 95.3%
Total: 10000 W: 4637 L: 4615 D: 748
Ptnml(0-2): 0, 75, 4828, 97, 0
nElo: 5.83 ± 6.8 (95%) PairsRatio: 1.29
closes https://github.com/official-stockfish/Stockfish/pull/6080
Bench: 2422771
2025-05-21 07:25:40 +02:00
Daniel Samek and Joost VandeVondele
ccfa651968
Remove full depth search reduction when cutNode
...
Passed STC-simplification bounds:
https://tests.stockfishchess.org/tests/view/6829dd6d6ec7634154f99fd3
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 67872 W: 17629 L: 17443 D: 32800
Ptnml(0-2): 167, 7988, 17451, 8152, 178
Passed LTC-simplification bounds:
https://tests.stockfishchess.org/tests/view/6829f2176ec7634154f9a01c
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 94818 W: 24328 L: 24184 D: 46306
Ptnml(0-2): 52, 10246, 26667, 10394, 50
closes https://github.com/official-stockfish/Stockfish/pull/6074
bench: 2245168
2025-05-19 20:44:14 +02:00
Shawn Xu and Joost VandeVondele
0f102f3692
Simplify Quiet Early Move Penalty
...
Passed STC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 185344 W: 47995 L: 47939 D: 89410
Ptnml(0-2): 527, 21898, 47754, 21978, 515
https://tests.stockfishchess.org/tests/view/682a47536ec7634154f9a3bc
Passed LTC:
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 101706 W: 26050 L: 25912 D: 49744
Ptnml(0-2): 53, 11056, 28499, 11190, 55
https://tests.stockfishchess.org/tests/view/682a61736ec7634154f9a50e
closes https://github.com/official-stockfish/Stockfish/pull/6072
Bench: 2012032
2025-05-19 07:45:28 +02:00
Shawn Xu and Joost VandeVondele
39942db3ff
Simplify In-Check Statscore
...
Passed Non-regression STC:
LLR: 2.98 (-2.94,2.94) <-1.75,0.25>
Total: 129760 W: 33701 L: 33580 D: 62479
Ptnml(0-2): 359, 15248, 33575, 15309, 389
https://tests.stockfishchess.org/tests/view/681a88193629b02d74b17123
Passed Non-regression LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 519612 W: 132224 L: 132512 D: 254876
Ptnml(0-2): 246, 56823, 145960, 56527, 250
https://tests.stockfishchess.org/tests/view/681f9ed43629b02d74b177c3
closes https://github.com/official-stockfish/Stockfish/pull/6070
bench: 2046462
2025-05-19 07:39:56 +02:00
Daniel Monroe and Joost VandeVondele
6e9b5af0f0
Check evaluation after ttMove before doing a tt cut
...
Passed STC
LLR: 2.97 (-2.94,2.94) <0.00,2.00>
Total: 239136 W: 62222 L: 61608 D: 115306
Ptnml(0-2): 675, 28046, 61525, 28634, 688
https://tests.stockfishchess.org/tests/view/681053293629b02d74b1668f
Passed LTC
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 448770 W: 115237 L: 114088 D: 219445
Ptnml(0-2): 177, 48128, 126619, 49291, 170
https://tests.stockfishchess.org/tests/view/681902de3629b02d74b16f6d
closes https://github.com/official-stockfish/Stockfish/pull/6069
Bench: 2035432
2025-05-19 07:37:15 +02:00
FauziAkram and Joost VandeVondele
4f76768fcf
Remove a moveCount condition
...
Passed STC:
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 70816 W: 18315 L: 18134 D: 34367
Ptnml(0-2): 210, 8213, 18360, 8436, 189
https://tests.stockfishchess.org/tests/view/68248197a527315e07cccb2d
Passed LTC:
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 121770 W: 31248 L: 31130 D: 59392
Ptnml(0-2): 61, 13338, 33995, 13404, 87
https://tests.stockfishchess.org/tests/view/68272ff46ec7634154f998ad
closes https://github.com/official-stockfish/Stockfish/pull/6067
bench: 2319161
2025-05-19 07:31:26 +02:00
Mapika and Joost VandeVondele
1b6975ac41
Add quiet move streak tracking to search stack
...
Passed STC:
LLR: 2.93 (-2.94,2.94) <0.00,2.00>
Total: 109344 W: 28473 L: 28053 D: 52818
Ptnml(0-2): 320, 12756, 28085, 13206, 305
https://tests.stockfishchess.org/tests/view/6828c43e6ec7634154f99a10
Passed LTC:
LLR: 2.96 (-2.94,2.94) <0.50,2.50>
Total: 76308 W: 19721 L: 19323 D: 37264
Ptnml(0-2): 39, 8145, 21386, 8547, 37
https://tests.stockfishchess.org/tests/view/6828f65a6ec7634154f99b72
closes https://github.com/official-stockfish/Stockfish/pull/6066
Bench: 2161814
2025-05-19 07:27:26 +02:00
Shawn Xu and Joost VandeVondele
6f445631ab
Simplify Futility Margin
...
Passed STC Non-regression:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 159008 W: 41500 L: 41414 D: 76094
Ptnml(0-2): 501, 18821, 40759, 18937, 486
https://tests.stockfishchess.org/tests/view/680ff9e23629b02d74b1663a
Passed LTC Non-regression:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 163572 W: 41617 L: 41543 D: 80412
Ptnml(0-2): 90, 17755, 46024, 17825, 92
https://tests.stockfishchess.org/tests/view/6814dd973629b02d74b16bac
closes https://github.com/official-stockfish/Stockfish/pull/6065
Bench: 2018775
2025-05-19 07:22:20 +02:00
Shawn Xu and Joost VandeVondele
6b7e05f0c5
Simplify PCM TTMove Bonus
...
Passed Non-regression STC:
LLR: 2.97 (-2.94,2.94) <-1.75,0.25>
Total: 114048 W: 29597 L: 29459 D: 54992
Ptnml(0-2): 315, 13619, 29045, 13703, 342
https://tests.stockfishchess.org/tests/view/681e83533629b02d74b17701
Passed Non-regression LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 61014 W: 15582 L: 15405 D: 30027
Ptnml(0-2): 25, 6485, 17307, 6668, 22
https://tests.stockfishchess.org/tests/view/68226b523629b02d74b17b89
closes https://github.com/official-stockfish/Stockfish/pull/6061
bench 2016566
2025-05-19 07:13:39 +02:00
FauziAkram and Disservin
c4e2479a75
Introducing a depth component to the penalty.
...
Passed STC:
LLR: 2.97 (-2.94,2.94) <0.00,2.00>
Total: 31648 W: 8358 L: 8050 D: 15240
Ptnml(0-2): 78, 3596, 8182, 3876, 92
https://tests.stockfishchess.org/tests/view/680fa73d3629b02d74b165a9
Passed LTC:
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 177720 W: 45524 L: 44920 D: 87276
Ptnml(0-2): 91, 19130, 49813, 19736, 90
https://tests.stockfishchess.org/tests/view/68109e2c3629b02d74b166ee
closes https://github.com/official-stockfish/Stockfish/pull/6060
Bench: 2251724
2025-05-13 20:47:50 +02:00
Shawn Xu and Disservin
b5f11085dd
Do more extensions in nodes far from the root
...
Parameters found by @FauziAkram in the latest tune.
Passed VVLTC w/ LTC Bound:
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 243264 W: 63452 L: 62774 D: 117038
Ptnml(0-2): 18, 22494, 75934, 23164, 22
https://tests.stockfishchess.org/tests/view/680e82b23629b02d74b15e27
Passed VVLTC w/ STC Bound:
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 37838 W: 9935 L: 9667 D: 18236
Ptnml(0-2): 6, 3383, 11873, 3651, 6
https://tests.stockfishchess.org/tests/view/681e707a3629b02d74b176e8
STC Elo Estimate:
Elo: -0.49 ± 2.4 (95%) LOS: 34.8%
Total: 20000 W: 5118 L: 5146 D: 9736
Ptnml(0-2): 55, 2365, 5182, 2349, 49
nElo: -0.96 ± 4.8 (95%) PairsRatio: 0.99
https://tests.stockfishchess.org/tests/view/6822af4b3629b02d74b17be6
closes https://github.com/official-stockfish/Stockfish/pull/6059
Bench: 2135382
2025-05-13 20:46:27 +02:00
Daniel Monroe and Disservin
1f9af9966f
Simplify futility pruning term
...
Passed simplification STC
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 49920 W: 12933 L: 12727 D: 24260
Ptnml(0-2): 141, 5865, 12752, 6051, 151
https://tests.stockfishchess.org/tests/view/681d01d33629b02d74b1756e
Passed simplification LTC
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 68808 W: 17573 L: 17402 D: 33833
Ptnml(0-2): 23, 7343, 19511, 7494, 33
https://tests.stockfishchess.org/tests/view/681e33843629b02d74b176b1
closes https://github.com/official-stockfish/Stockfish/pull/6058
Bench: 2041086
2025-05-13 20:42:36 +02:00
FauziAkram and Disservin
40ef7b1212
Simplify probcut
...
Passed STC:
LLR: 2.97 (-2.94,2.94) <-1.75,0.25>
Total: 80800 W: 20947 L: 20774 D: 39079
Ptnml(0-2): 217, 9446, 20906, 9609, 222
https://tests.stockfishchess.org/tests/view/680e83163629b02d74b15e2a
Passed LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 359004 W: 91362 L: 91486 D: 176156
Ptnml(0-2): 177, 39133, 101007, 39007, 178
https://tests.stockfishchess.org/tests/view/680e95db3629b02d74b15e7a
closes https://github.com/official-stockfish/Stockfish/pull/6054
Bench: 2060860
2025-05-13 20:41:01 +02:00
FauziAkram and Disservin
d4b405a5a6
Remove a cutNode condition
...
Allow some nodes to spawn even deeper lmr searches, also for cutNodes
Passed STC:
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 59072 W: 15387 L: 15190 D: 28495
Ptnml(0-2): 145, 6956, 15166, 7095, 174
https://tests.stockfishchess.org/tests/view/6810b4c13629b02d74b16714
Passed LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 93294 W: 23737 L: 23591 D: 45966
Ptnml(0-2): 47, 10116, 26169, 10274, 41
https://tests.stockfishchess.org/tests/view/681170613629b02d74b167f3
closes https://github.com/official-stockfish/Stockfish/pull/6052
Bench: 1937565
2025-05-13 20:39:23 +02:00
Shawn Xu and Disservin
e9925b122f
Simplify ttCapture LMR
...
Passed Non-regression STC:
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 51104 W: 13389 L: 13184 D: 24531
Ptnml(0-2): 182, 5940, 13068, 6215, 147
https://tests.stockfishchess.org/tests/view/680ef2503629b02d74b16498
Passed Non-regression LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 73350 W: 18804 L: 18638 D: 35908
Ptnml(0-2): 30, 7906, 20639, 8068, 32
https://tests.stockfishchess.org/tests/view/6810510e3629b02d74b1668a
closes https://github.com/official-stockfish/Stockfish/pull/6044
Bench: 1805151
2025-05-13 20:36:42 +02:00
Michael Chaly and Disservin
b73c8982df
Another scaling revert
...
Revert recent passer because it seems to not scale for longer time
controls. Adjust comments accordingly.
Passed VVLTC SPRT with STC bounds:
https://tests.stockfishchess.org/tests/view/680ddff73629b02d74b15b3f
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 198316 W: 51317 L: 50846 D: 96153
Ptnml(0-2): 17, 18459, 61737, 18926, 19
Passed VVLTC SPRT with LTC bounds:
https://tests.stockfishchess.org/tests/view/680d5b7e3629b02d74b15a21
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 123274 W: 31738 L: 31283 D: 60253
Ptnml(0-2): 7, 11444, 38282, 11895, 9
closes https://github.com/official-stockfish/Stockfish/pull/6038
Bench: 2173845
2025-05-13 20:32:54 +02:00
Shawn Xu and Disservin
81cc004060
Remove risk tolerance
...
Passed STC:
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 379328 W: 97567 L: 97724 D: 184037
Ptnml(0-2): 909, 44861, 98314, 44638, 942
https://tests.stockfishchess.org/tests/view/680defc63629b02d74b15b62
Passed LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 160752 W: 40762 L: 40685 D: 79305
Ptnml(0-2): 60, 17548, 45091, 17609, 68
https://tests.stockfishchess.org/tests/view/680e8ff43629b02d74b15e65
closes https://github.com/official-stockfish/Stockfish/pull/6037
Bench: 1897340
2025-05-13 20:31:37 +02:00
ed6b8d179a
Refactor futility_margin
...
a small term to the futility calculation that depends on eval - beta.
Passed STC:
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 203136 W: 52797 L: 52239 D: 98100
Ptnml(0-2): 549, 23827, 52255, 24391, 546
https://tests.stockfishchess.org/tests/view/680e84a43629b02d74b15e2e
Passed LTC:
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 100356 W: 25950 L: 25507 D: 48899
Ptnml(0-2): 35, 10683, 28302, 11120, 38
https://tests.stockfishchess.org/tests/view/680ebcb03629b02d74b16040
closes https://github.com/official-stockfish/Stockfish/pull/6009
closes https://github.com/official-stockfish/Stockfish/pull/6041
Bench: 1815939
Co-authored-by: Michael Chaly <Vizvezdenec@gmail.com >
Co-authored-by: xu-shawn <50402888+xu-shawn@users.noreply.github.com >
2025-05-13 20:26:28 +02:00
pb00067 and Disservin
0f905b4e88
Preserve all moves in movepicker
...
Simplifies method otherPieceTypesMobile quite a bit and makes it more
precise. More precise because capturesSearched list not always contains
all processed captures:
although extremely rare, it can happen that a 'good' capture get pruned
at step 14 and doesn't make it to capturesSearched. (functional change
at higher depths)
passed STC-simplification bounds
https://tests.stockfishchess.org/tests/view/68025974cd501869c6698153
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 273664 W: 15658 L: 15681 D: 242325
Ptnml(0-2): 166, 10368, 115802, 10315, 181
passed LTC-simplification bounds
https://tests.stockfishchess.org/tests/view/6804b86acd501869c6698673
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 96780 W: 24547 L: 24419 D: 47814
Ptnml(0-2): 30, 8466, 31286, 8562, 46
Applied changes requested by disservin & retested to be on the safe side
STC:
https://tests.stockfishchess.org/tests/view/6806110698cd372e3aea5919
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 107392 W: 27739 L: 27606 D: 52047
Ptnml(0-2): 266, 10867, 31306, 10982, 275
LTC:
https://tests.stockfishchess.org/tests/view/6806346198cd372e3aea5965
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 233484 W: 59106 L: 59103 D: 115275
Ptnml(0-2): 116, 22787, 70939, 22778, 122
closes https://github.com/official-stockfish/Stockfish/pull/6005
Bench: 1857323
2025-05-13 20:23:43 +02:00
FauziAkram and Disservin
94e6c0498f
VVLTC Tune
...
Passed VVLTC with LTC bounds:
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 12800 W: 3432 L: 3184 D: 6184
Ptnml(0-2): 1, 1098, 3954, 1346, 1
https://tests.stockfishchess.org/tests/view/680e255e3629b02d74b15d5e
Passed VVLTC with STC bounds:
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 14402 W: 3865 L: 3625 D: 6912
Ptnml(0-2): 0, 1236, 4490, 1474, 1
https://tests.stockfishchess.org/tests/view/680e4dfb3629b02d74b15da6
Passed VVLTC third test (removing an unrelated change):
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 25584 W: 6670 L: 6398 D: 12516
Ptnml(0-2): 4, 2290, 7932, 2562, 4
https://tests.stockfishchess.org/tests/view/680e74223629b02d74b15def
closes https://github.com/official-stockfish/Stockfish/pull/6036
Bench: 1857323
2025-04-27 20:47:47 +02:00
Daniel Monroe and Disservin
af3692b2d0
Simplify second probcut to linear function of depth
...
Passed non-regression STC
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 57472 W: 14962 L: 14765 D: 27745
Ptnml(0-2): 140, 6715, 14817, 6936, 128
https://tests.stockfishchess.org/tests/view/680df1063629b02d74b15b69
Passed non-regression LTC
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 88416 W: 22499 L: 22348 D: 43569
Ptnml(0-2): 31, 9565, 24874, 9698, 40
https://tests.stockfishchess.org/tests/view/680df3a93629b02d74b15b7d
closes https://github.com/official-stockfish/Stockfish/pull/6035
Bench: 1792000
2025-04-27 20:31:40 +02:00
Michael Chaly and Disservin
3e26d3acc7
Do more pruning in moves loop
...
Effectively reverts one commit from some months ago.
Passed VVLTC SPRT with STC bounds
https://tests.stockfishchess.org/tests/view/680d39373629b02d74b156d7
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 405058 W: 104843 L: 104111 D: 196104
Ptnml(0-2): 35, 38029, 125672, 38755, 38
Passed VVLTC SPRT with LTC bounds
https://tests.stockfishchess.org/tests/view/680d1a3b3629b02d74b1563d
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 57032 W: 14917 L: 14588 D: 27527
Ptnml(0-2): 6, 5202, 17768, 5537, 3
closes https://github.com/official-stockfish/Stockfish/pull/6031
Bench: 1643819
2025-04-27 20:28:22 +02:00
FauziAkram and Disservin
267fd8a3d5
Tweak History Bonus
...
Inspired by @Ergodice , who came up first with the idea.
Passed STC:
LLR: 2.96 (-2.94,2.94) <0.00,2.00>
Total: 18400 W: 4867 L: 4576 D: 8957
Ptnml(0-2): 52, 2052, 4714, 2317, 65
https://tests.stockfishchess.org/tests/view/68062a3c98cd372e3aea5959
Passed LTC:
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 454338 W: 116461 L: 115294 D: 222583
Ptnml(0-2): 198, 49139, 127346, 50270, 216
https://tests.stockfishchess.org/tests/view/6806347c98cd372e3aea5967
Passed VLTC non-reg:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 385970 W: 98401 L: 98546 D: 189023
Ptnml(0-2): 51, 38958, 115105, 38827, 44
https://tests.stockfishchess.org/tests/view/680cfe873629b02d74b155cf
closes https://github.com/official-stockfish/Stockfish/pull/6030
Bench: 1715817
2025-04-27 20:14:38 +02:00
FauziAkram and Disservin
f98c178960
Improve quiet moves bonus
...
Inspired by an old test by Peregrine.
Passed STC:
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 453024 W: 117244 L: 116316 D: 219464
Ptnml(0-2): 1336, 53355, 116258, 54171, 1392
https://tests.stockfishchess.org/tests/view/680ccacc3629b02d74b15532
Passed LTC:
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 140550 W: 35990 L: 35462 D: 69098
Ptnml(0-2): 65, 15152, 39319, 15668, 71
https://tests.stockfishchess.org/tests/view/680d2ed73629b02d74b15691
closes https://github.com/official-stockfish/Stockfish/pull/6028
Bench: 1708152
2025-04-27 19:45:37 +02:00
Myself and Disservin
37cc2293ef
Replace complex probcut function with a precomputed table.
...
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/680de2683629b02d74b15b46
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 116000 W: 29864 L: 29742 D: 56394
Ptnml(0-2): 215, 11811, 33854, 11877, 243
closes https://github.com/official-stockfish/Stockfish/pull/6026
No functional change
2025-04-27 19:45:00 +02:00
Shawn Xu and Disservin
b0a7a34d3f
Simplify malus calculation
...
closes https://github.com/official-stockfish/Stockfish/pull/6024
No functional change
2025-04-27 19:43:47 +02:00
Shawn Xu and Disservin
4e49f8dff9
Clean up search
...
* Correct IIR scaling comments
* Replace `(PvNode || cutNode)` with `!allNode`
* Consistent formatting for scaler tags
* Add comments to some recently-introduced LMR terms
* Add comments on PCM bonus tweaks
Passed Non-regression STC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 389472 W: 102457 L: 102622 D: 184393
Ptnml(0-2): 1676, 41887, 107798, 41676, 1699
https://tests.stockfishchess.org/tests/view/67a0ea670774dfd78deb23cd
closes https://github.com/official-stockfish/Stockfish/pull/5854
Bench: 1585741
2025-04-27 19:26:02 +02:00
Michael Chaly and Joost VandeVondele
27428a61c2
Allow some nodes to spawn even deeper lmr searches
...
This includes nodes that were PvNode on a previous ply and only for non cutNodes and movecounts < 8.
Passed STC:
https://tests.stockfishchess.org/tests/view/680cdf2e3629b02d74b15576
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 123552 W: 31979 L: 31539 D: 60034
Ptnml(0-2): 322, 14449, 31803, 14871, 331
Passed LTC:
https://tests.stockfishchess.org/tests/view/680cf09a3629b02d74b15599
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 114306 W: 29310 L: 28837 D: 56159
Ptnml(0-2): 51, 12247, 32090, 12708, 57
closes https://github.com/official-stockfish/Stockfish/pull/6022
bench: 1585741
2025-04-26 22:40:41 +02:00
Michael Chaly and Joost VandeVondele
4b58079485
Simplify and cleanup futility pruning for child nodes
...
This patch removes (eval - beta) / 8 addition and adjusts constants
accordingly, also moves every calculation into futility_margin function.
Passed STC:
https://tests.stockfishchess.org/tests/view/6806d00f878abf56f9a0d524
LLR: 2.99 (-2.94,2.94) <-1.75,0.25>
Total: 483456 W: 124592 L: 124860 D: 234004
Ptnml(0-2): 1419, 57640, 123927, 57274, 1468
Passed LTC:
https://tests.stockfishchess.org/tests/view/680cceb33629b02d74b1554c
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 263868 W: 67076 L: 67104 D: 129688
Ptnml(0-2): 155, 28893, 73846, 28905, 135
closes https://github.com/official-stockfish/Stockfish/pull/6021
bench: 1618439
2025-04-26 22:38:14 +02:00
breatn and Joost VandeVondele
f590767b91
Adaptive beta cut
...
passed STC:
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 197088 W: 51084 L: 50533 D: 95471
Ptnml(0-2): 547, 23201, 50577, 23592, 627
https://tests.stockfishchess.org/tests/view/680604d798cd372e3aea58fe
passed LTC
LLR: 2.96 (-2.94,2.94) <0.50,2.50>
Total: 127950 W: 32719 L: 32214 D: 63017
Ptnml(0-2): 69, 13825, 35673, 14348, 60
https://tests.stockfishchess.org/tests/view/6805eae498cd372e3aea588a
closes https://github.com/official-stockfish/Stockfish/pull/6012
bench 1579003
2025-04-26 22:23:26 +02:00
Nonlinear2 and Joost VandeVondele
0dcfe096d6
Increase full depth search reduction when cutNode
...
In addition to the core patch, improve the use of `isTTMove`:
- this name was used to mean both `bestMove == ttData.move` and `move == ttData.move`, so i replaced the argument `isTTMove` of `update_all_stats` with `TTMove` directly.
- `ttData.move == move` was still used in some places instead of `ss->isTTMove`. I replaced these to be more consistent.
Passed STC:
https://tests.stockfishchess.org/tests/view/68057b8f98cd372e3aea3472
LLR: 2.93 (-2.94,2.94) <0.00,2.00>
Total: 38400 W: 10048 L: 9734 D: 18618
Ptnml(0-2): 102, 4360, 9956, 4686, 96
Passed LTC:
https://tests.stockfishchess.org/tests/view/68057f7c98cd372e3aea3842
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 312666 W: 79494 L: 78616 D: 154556
Ptnml(0-2): 144, 33809, 87563, 34659, 158
closes https://github.com/official-stockfish/Stockfish/pull/6007
Bench: 1623376
2025-04-26 22:11:40 +02:00
Daniel Monroe and Joost VandeVondele
88a524c552
Tweak futility formula
...
Passed STC
LLR: 2.93 (-2.94,2.94) <0.00,2.00>
Total: 248448 W: 64344 L: 63718 D: 120386
Ptnml(0-2): 750, 29172, 63783, 29740, 779
https://tests.stockfishchess.org/tests/view/68056f5598cd372e3aea2901
Passed LTC
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 118824 W: 30358 L: 29874 D: 58592
Ptnml(0-2): 59, 12797, 33228, 13257, 71
https://tests.stockfishchess.org/tests/view/6805675698cd372e3aea20d0
closes https://github.com/official-stockfish/Stockfish/pull/6004
bench 1839796
2025-04-26 22:06:18 +02:00
FauziAkram and Joost VandeVondele
f6b0d53a99
Re-adding the 5th continuation history
...
the two simplifications (#5992 and #5978 ) are strongly correlated.
Passed STC:
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 98016 W: 25415 L: 25008 D: 47593
Ptnml(0-2): 291, 11509, 25005, 11908, 295
https://tests.stockfishchess.org/tests/view/6802774fcd501869c6698168
Passed LTC:
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 588588 W: 150073 L: 148633 D: 289882
Ptnml(0-2): 281, 63615, 165078, 65023, 297
https://tests.stockfishchess.org/tests/view/6804cbdacd501869c66987dc
closes https://github.com/official-stockfish/Stockfish/pull/6002
Bench: 1783315
2025-04-26 21:58:46 +02:00
Michael Chaly and Joost VandeVondele
449a8b017e
Do second step of shallower search
...
Specifically if lmr depth is not lower than new depth and value is really bad.
Passed STC:
https://tests.stockfishchess.org/tests/view/6805444898cd372e3aea0494
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 77664 W: 20136 L: 19762 D: 37766
Ptnml(0-2): 214, 9006, 20039, 9338, 235
Passed LTC:
https://tests.stockfishchess.org/tests/view/680549b298cd372e3aea05ac
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 115458 W: 29447 L: 28971 D: 57040
Ptnml(0-2): 62, 12357, 32421, 12821, 68
closes https://github.com/official-stockfish/Stockfish/pull/6001
bench: 1515501
2025-04-26 21:52:36 +02:00
Shawn Xu and Joost VandeVondele
4176ad7b0a
simplify risk tolerance
...
Passed Non-regression STC:
LLR: 2.98 (-2.94,2.94) <-1.75,0.25>
Total: 73408 W: 19028 L: 18844 D: 35536
Ptnml(0-2): 201, 8709, 18743, 8807, 244
https://tests.stockfishchess.org/tests/view/68051f3698cd372e3ae9f63a
Passed Non-regression LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 91236 W: 23193 L: 23045 D: 44998
Ptnml(0-2): 34, 9908, 25599, 10030, 47
https://tests.stockfishchess.org/tests/view/6805239498cd372e3ae9fa41
closes https://github.com/official-stockfish/Stockfish/pull/6000
bench 1864632
2025-04-26 21:49:38 +02:00
Shawn Xu and Joost VandeVondele
16cd38dba1
Tweak TT Move Reduction by TT Move History
...
Passed STC:
LLR: 2.93 (-2.94,2.94) <0.00,2.00>
Total: 62496 W: 16197 L: 15844 D: 30455
Ptnml(0-2): 200, 7234, 16011, 7619, 184
https://tests.stockfishchess.org/tests/view/67f7fa9b31d7cf8afdc4609c
Passed LTC:
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 400470 W: 102068 L: 101012 D: 197390
Ptnml(0-2): 201, 43207, 112347, 44295, 185
https://tests.stockfishchess.org/tests/view/67f927b0cd501869c66975e0
Passed VVLTC Non-regression:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 125394 W: 32408 L: 32304 D: 60682
Ptnml(0-2): 3, 11702, 39188, 11796, 8
https://tests.stockfishchess.org/tests/view/6804c215cd501869c66986b9
closes https://github.com/official-stockfish/Stockfish/pull/5998
bench 1760988
2025-04-26 21:43:11 +02:00
Shawn Xu and Joost VandeVondele
f273eea71f
Remove non-functional accumulator reset
...
Passed Non-regression STC:
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 219360 W: 56600 L: 56583 D: 106177
Ptnml(0-2): 582, 23419, 61620, 23518, 541
https://tests.stockfishchess.org/tests/view/67fad20dcd501869c669780f
closes https://github.com/official-stockfish/Stockfish/pull/5986
no functional change
2025-04-26 21:30:01 +02:00
Daniel Monroe and Disservin
3d18ad719b
Skip 5th continuation history
...
Passed simplification STC
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 70208 W: 18098 L: 17914 D: 34196
Ptnml(0-2): 199, 8300, 17907, 8514, 184
https://tests.stockfishchess.org/tests/view/67ed889b31d7cf8afdc451cb
Passed simplification LTC
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 485004 W: 122703 L: 122956 D: 239345
Ptnml(0-2): 288, 53162, 135805, 53009, 238
https://tests.stockfishchess.org/tests/view/67ee810231d7cf8afdc452ea
closes https://github.com/official-stockfish/Stockfish/pull/5992
Bench: 1715901
2025-04-18 14:32:26 +02:00
Guenther Demetz and Disservin
698c069bba
Move node increment inside do_move function
...
Move node increment inside do_move function so that we can centralize
the increment into a single point.
closes https://github.com/official-stockfish/Stockfish/pull/5989
No functional change
2025-04-18 14:32:26 +02:00
Shawn Xu and Disservin
2b4926e091
Simplify TT Move History
...
Part 1 passed Non-regression STC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 195552 W: 50394 L: 50349 D: 94809
Ptnml(0-2): 581, 23222, 50122, 23273, 578
https://tests.stockfishchess.org/tests/view/67eb6ea831d7cf8afdc44c30
Part 2 passed Non-regression STC:
LLR: 2.92 (-2.94,2.94) <-1.75,0.25>
Total: 181664 W: 46786 L: 46727 D: 88151
Ptnml(0-2): 517, 21403, 46974, 21380, 558
https://tests.stockfishchess.org/tests/view/67eb6f3331d7cf8afdc44c33
Passed Non-regression LTC:
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 155454 W: 39496 L: 39412 D: 76546
Ptnml(0-2): 77, 16950, 43580, 17052, 68
https://tests.stockfishchess.org/tests/view/67eee76531d7cf8afdc45358
Passed Non-regression VLTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 118266 W: 30263 L: 30148 D: 57855
Ptnml(0-2): 11, 11844, 35309, 11957, 12
https://tests.stockfishchess.org/tests/view/67f2414a31d7cf8afdc45760
closes https://github.com/official-stockfish/Stockfish/pull/5987
Bench: 1792850
2025-04-18 14:32:26 +02:00
pb00067 and Disservin
d2d046c2a4
Improve stalemate detection during search
...
Currently SF’s quiescence search like most alpha-beta based engines
doesn’t verify for stalemate because doing it each leaf position is to
expensive and costs elo. However in certain positions this creates a
blindspot for SF, not recognizing soon enough that the opponent can
reach a stalemate by sacrifycing his last mobile heavy piece(s). This
tactical motif & it’s measure are similar to zugzwang & verification
search: the measure itself does not gain elo, but prevents SF from
loosing/drawing games in an awkward way.
The fix consists of 3 measures:
1. Make qsearch verify for stalemate on transitions to pure KP-material
for the side to move with our last Rook/Queen just been captured. In
fact this is the scenario where stalemate happens with highest
frequency. The stalemate-verification itself is optimized by merely
checking for pawn pushes & king mobility (captures were already
tried by qssearch)
2. Another culprit for the issue figured out to be SEE based pruning for
checks in step 14. Here often the move forcing the stalemate (or
forcing the opponent to not retake) get pruned away and it need to
much time to reach enough depth. To encounter this we verify
following conditions:
- a) side to move is happy with a draw (alpha < 0)
- b) we are about to sacrify our last heavy & unique mobile piece in
this position.
- c) this piece doesn’t move away from our kingring giving the king a
new square to move.
When all 3 conditions meet we don’t prune the move, because there
is a good chance that capturing the piece means stalemate.
3. Store terminal nodes (mates & stalemates) in TT with higher depth
than searched depth. This prevents SF from:
- reanalyzing the node (=trying to generate legal moves) in vain at
each iterative deepening step.
- overwriting an already correct draw-evaluation from a previous shallow
normal search by a qsearch which doesn’t recognize stalemate and might
store a verry erratic evaluation.
This is due to the 4 constant in the TT-overwrite condition: d -
DEPTH_ENTRY_OFFSET + 2 * pv > depth8 – 4 which allows qs to override
entries made by normal searches with depth <=4.
This 3hrd measure however is not essential for fixing the issue, but
tests (one of vdv & one of mine) seem to suggest that this measure
brings some small benefit.
Another other position where SF benefits from this fix is for instance
Position FEN 8/8/8/1B6/6p1/8/3K1Ppp/3N2kr w - - 0 1 bm f4 +M9
P.S.: Also this issue higly depends on the used net, how good the net is
at evaluate such mobility restricted positions. SF16 was pretty good in
solving 2rr4/5pBk/PqP3p1/1N3pPp/1PQ1bP1P/8/3R4/R4K2 b - - 0 40 bm Rxc6
(< 1 second) while SF16_1 with introduction of the dual net needs about
1,5 minutes and SF17.1 requires 3 minutes to find the drawing move Rxc6.
P.S.2: Using more threads produces indeterminism & using high hash
pressure makes SF reevaluate explored positions more often which makes
it more likely to solve the position. To have stable meaningful results
I tested therfore with one single thread and low hash pressure.
Preliminary LTC test at 30k games
https://tests.stockfishchess.org/tests/view/67ece7a931d7cf8afdc44e18 Elo: 0.04 ± 2.0 (95%) LOS: 51.7%
Total: 24416 W: 6226 L: 6223 D: 11967
Ptnml(0-2): 12, 2497, 7185, 2504, 10
nElo: 0.09 ± 4.4 (95%) PairsRatio: 1.00
Passed LTC no-regression sprt
https://tests.stockfishchess.org/tests/view/67ee8e4631d7cf8afdc452fb
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 401556 W: 101612 L: 101776 D: 198168
Ptnml(0-2): 152, 42241, 116170, 42049, 166
closes https://github.com/official-stockfish/Stockfish/pull/5983
fixes https://github.com/official-stockfish/Stockfish/issues/5899
Bench: 1721673
2025-04-18 14:32:26 +02:00
FauziAkram and Disservin
44efbaddea
Simplify bonusScale calculation
...
Allowing this specific term to potentially become negative for low depth
values is ok, because the overall `bonusScale` is explicitly ensured to
be non-negative a few lines later: bonusScale = std::max(bonusScale, 0);
Passed STC:
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 164928 W: 42446 L: 42368 D: 80114
Ptnml(0-2): 497, 19551, 42306, 19597, 513
https://tests.stockfishchess.org/tests/view/67debf0b8888403457d8736c
Passed LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 234942 W: 59539 L: 59537 D: 115866
Ptnml(0-2): 113, 25639, 65964, 25643, 112
https://tests.stockfishchess.org/tests/view/67e2e1c48888403457d87768
closes https://github.com/official-stockfish/Stockfish/pull/5979
Bench: 1933843
2025-04-18 14:32:26 +02:00
FauziAkram and Disservin
cf8b3637a0
Improve futility pruning
...
Adding a small term to the futility calculation that depends on eval - beta.
Refactored to a simpler form.
Passed STC:
LLR: 2.93 (-2.94,2.94) <0.00,2.00>
Total: 74176 W: 19323 L: 18954 D: 35899
Ptnml(0-2): 226, 8576, 19117, 8941, 228
https://tests.stockfishchess.org/tests/view/67e6b0946682f97da2178eaf
Passed LTC:
LLR: 2.96 (-2.94,2.94) <0.50,2.50>
Total: 135090 W: 34499 L: 33983 D: 66608
Ptnml(0-2): 79, 14403, 38040, 14969, 54
https://tests.stockfishchess.org/tests/view/67e757cc6682f97da2178f62
closes https://github.com/official-stockfish/Stockfish/pull/5970
Bench: 1875196
2025-04-18 14:32:19 +02:00
Daniel Monroe and Disservin
fb6a3e04ec
Simply use non_pawn_material rather than summing tuned terms
...
Passed simplification STC
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 136576 W: 35285 L: 35175 D: 66116
Ptnml(0-2): 410, 16179, 34997, 16295, 407
https://tests.stockfishchess.org/tests/view/67e5dc736682f97da2178da6
Passed rebased simplification LTC
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 85482 W: 21812 L: 21658 D: 42012
Ptnml(0-2): 34, 9260, 24022, 9368, 57
https://tests.stockfishchess.org/tests/view/67e852cb31d7cf8afdc44966
closes https://github.com/official-stockfish/Stockfish/pull/5965
Bench: 2006483
2025-04-18 14:32:19 +02:00
Shawn Xu and Disservin
d7c04a9429
Introduce TT Move History Double Extensions
...
Passed VVLTC w/ STC bounds:
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 74918 W: 19436 L: 19120 D: 36362
Ptnml(0-2): 6, 6890, 23354, 7200, 9
https://tests.stockfishchess.org/tests/view/67e4a1088888403457d878bb
Passed VVLTC w/ LTC bounds:
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 111706 W: 29050 L: 28619 D: 54037
Ptnml(0-2): 13, 10218, 34959, 10651, 12
https://tests.stockfishchess.org/tests/view/67d6877c517865b4a2dfd36b
STC Elo Estimate:
Elo: 1.26 ± 2.0 (95%) LOS: 88.8%
Total: 30000 W: 7855 L: 7746 D: 14399
Ptnml(0-2): 104, 3531, 7630, 3622, 113
nElo: 2.44 ± 3.9 (95%) PairsRatio: 1.03
https://tests.stockfishchess.org/tests/view/67eacb8c31d7cf8afdc44b99
closes https://github.com/official-stockfish/Stockfish/pull/5961
Bench: 1887897
2025-04-02 17:53:05 +02:00
Daniel Samek and Disservin
d942e13398
Less fail high cnt in the condition
...
Passed STC:
https://tests.stockfishchess.org/tests/view/67e027538888403457d87535
LLR: 2.93 (-2.94,2.94) <0.00,2.00>
Total: 164000 W: 42535 L: 42034 D: 79431
Ptnml(0-2): 478, 19228, 42113, 19677, 504
Passed LTC:
https://tests.stockfishchess.org/tests/view/67e3c21b8888403457d87808
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 139176 W: 35500 L: 34975 D: 68701
Ptnml(0-2): 54, 15038, 38898, 15525, 73
closes https://github.com/official-stockfish/Stockfish/pull/5960
Bench: 1921404
2025-04-02 17:52:38 +02:00
