Linmiao Xu
025da6a0d1
Give positional output more weight in nnue eval
...
This effectively reverts the removal of delta in:
https://github.com/official-stockfish/Stockfish/pull/5373
Passed STC:
https://tests.stockfishchess.org/tests/view/6664d41922234461cef58e6b
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 56448 W: 14849 L: 14500 D: 27099
Ptnml(0-2): 227, 6481, 14457, 6834, 225
Passed LTC:
https://tests.stockfishchess.org/tests/view/666587a1996b40829f4ee007
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 91686 W: 23402 L: 22963 D: 45321
Ptnml(0-2): 78, 10205, 24840, 10640, 80
closes https://github.com/official-stockfish/Stockfish/pull/5382
bench 1160467
2024-06-12 09:17:04 +02:00
Dubslow
7d4ffa175c
Remove delta from evaluation
...
Passed STC: https://tests.stockfishchess.org/tests/view/6660e49c6489614cdad14e29
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 188768 W: 48907 L: 48854 D: 91007
Ptnml(0-2): 584, 22571, 48005, 22656, 568
Passed LTC: https://tests.stockfishchess.org/tests/view/6660ff9791e372763104b38c
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 310680 W: 78651 L: 78727 D: 153302
Ptnml(0-2): 180, 34818, 85433, 34716, 193
closes https://github.com/official-stockfish/Stockfish/pull/5373
Bench: 1214575
2024-06-08 23:34:03 +02:00
cj5716
5688b188cc
Simplify evaluation constants
...
Passed STC (<0, 2> by accident):
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 346016 W: 89529 L: 88756 D: 167731
Ptnml(0-2): 1012, 41074, 88027, 41919, 976
https://tests.stockfishchess.org/tests/view/6659d6ecf426908fcc6b6929
Passed LTC:
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 89862 W: 22887 L: 22734 D: 44241
Ptnml(0-2): 45, 9999, 24694, 10144, 49
https://tests.stockfishchess.org/tests/view/665a6ebb062b2c3cf814fde8
Passed LTC (Rebased):
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 325500 W: 82734 L: 82826 D: 159940
Ptnml(0-2): 193, 36409, 89665, 36263, 220
https://tests.stockfishchess.org/tests/view/665bd39f44e8416a9cdc1909
closes https://github.com/official-stockfish/Stockfish/pull/5361
Bench 961982
2024-06-05 21:49:29 +02:00
Gahtan Nahdi
daaccd9fc9
Simplify smallnet threshold
...
remove pawncount
Passed STC non-reg:
https://tests.stockfishchess.org/tests/view/665e4548fd45fb0f907c80d5
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 60896 W: 15710 L: 15518 D: 29668
Ptnml(0-2): 149, 7145, 15660, 7353, 141
Passed LTC non-reg:
https://tests.stockfishchess.org/tests/view/665e4c52fd45fb0f907c815f
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 58068 W: 14773 L: 14590 D: 28705
Ptnml(0-2): 16, 6368, 16090, 6537, 23
closes https://github.com/official-stockfish/Stockfish/pull/5349
Bench: 1343156
2024-06-05 20:59:56 +02:00
Dubslow
86b564055d
Remove delta, adjusted, complexity from nnue code
...
...rather they're the consumer's concern whether to tweak the result or not.
Passed STC:
https://tests.stockfishchess.org/tests/view/665cea9ffd45fb0f907c53bd
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 69696 W: 18101 L: 17918 D: 33677
Ptnml(0-2): 195, 8171, 17929, 8362, 191
Passed LTC:
https://tests.stockfishchess.org/tests/view/665cf761fd45fb0f907c5406
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 63720 W: 16344 L: 16165 D: 31211
Ptnml(0-2): 32, 6990, 17625, 7193, 20
Non functional except for rounding issues of OutputScale changing bench.
closes https://github.com/official-stockfish/Stockfish/pull/5344
Bench: 1378596
2024-06-03 23:27:58 +02:00
Gahtan Nahdi
fe298953f8
Simplify smallnet threshold
...
Turns the quadratic threshold to a linear one
STC non-reg:
https://tests.stockfishchess.org/tests/view/665ba0b744e8416a9cdc188d
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 330432 W: 85351 L: 85454 D: 159627
Ptnml(0-2): 888, 39643, 84283, 39488, 914
LTC non-reg:
https://tests.stockfishchess.org/tests/view/665cd60ffd45fb0f907c4306
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 139146 W: 35194 L: 35093 D: 68859
Ptnml(0-2): 58, 15523, 38313, 15618, 61
closes https://github.com/official-stockfish/Stockfish/pull/5342
Bench: 1057383
2024-06-03 23:18:36 +02:00
MinetaS
180cab4438
Simplify 50 move rule dampening
...
Refactor the logic of 50 move rule dampening by removing a constant.
Passed non-regression STC:
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 35232 W: 9214 L: 8992 D: 17026
Ptnml(0-2): 114, 4081, 8999, 4313, 109
https://tests.stockfishchess.org/tests/view/665a329013d08af3c1725610
Passed non-regression LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 38406 W: 9732 L: 9530 D: 19144
Ptnml(0-2): 14, 4132, 10708, 4336, 13
https://tests.stockfishchess.org/tests/view/665a370913d08af3c1725651
https://github.com/official-stockfish/Stockfish/pull/5327
Bench: 1059739
2024-06-01 20:09:47 +02:00
Linmiao Xu
0ef809ac71
Quadratic smallnet threshold with re-evaluation
...
The threshold now decreases more quickly as pawn count decreases,
using the smallnet more compared to before.
Combo of two eval patches:
https://tests.stockfishchess.org/tests/view/66576c5f6b0e318cefa8d26e
https://tests.stockfishchess.org/tests/view/664ced40830eb9f886616a77
Passed STC:
https://tests.stockfishchess.org/tests/view/66588c136b0e318cefa8ff21
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 112608 W: 29336 L: 28908 D: 54364
Ptnml(0-2): 344, 13223, 28718, 13699, 320
Passed LTC:
https://tests.stockfishchess.org/tests/view/6658c8786b0e318cefa900f5
LLR: 2.96 (-2.94,2.94) <0.50,2.50>
Total: 108288 W: 27493 L: 27026 D: 53769
Ptnml(0-2): 54, 11821, 29930, 12282, 57
closes https://github.com/official-stockfish/Stockfish/pull/5323
bench 1728074
2024-06-01 19:56:05 +02:00
Linmiao Xu
5ab3fe6db8
Simplify blending eval with nnue complexity
...
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/66567377a86388d5e27dd89c
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 144000 W: 37443 L: 37338 D: 69219
Ptnml(0-2): 587, 17260, 36208, 17351, 594
Passed non-regression LTC:
https://tests.stockfishchess.org/tests/view/66567f29a86388d5e27dd924
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 112326 W: 28550 L: 28421 D: 55355
Ptnml(0-2): 66, 12732, 30434, 12869, 62
closes https://github.com/official-stockfish/Stockfish/pull/5305
bench 1554486
2024-05-30 14:18:39 +02:00
Stéphane Nicolet
c7b80f6c8a
Merge pawn count terms using their average
...
This simplification patch merges the pawn count terms in the eval
formula with the material term, updating the offset constant for
the nnue part of the formula from 34000 to 34300 because the average
pawn count in middlegame positions evaluated during search is around 8.
STC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 138240 W: 35834 L: 35723 D: 66683
Ptnml(0-2): 527, 16587, 34817, 16626, 563
https://tests.stockfishchess.org/tests/view/6653f474a86388d5e27daaac
LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 454272 W: 114787 L: 115012 D: 224473
Ptnml(0-2): 246, 51168, 124553, 50903, 266
https://tests.stockfishchess.org/tests/view/6654f256a86388d5e27db131
closes https://github.com/official-stockfish/Stockfish/pull/5303
Bench: 1279635
2024-05-29 18:57:32 +02:00
Linmiao Xu
8bc3fd3871
Lower smallnet threshold with tuned eval params
...
The smallnet threshold is now below the training data range
of the current smallnet (simple eval diff > 1k, nn-baff1edelf90.nnue)
when no pawns are on the board.
Params found with spsa at 93k / 120k games at 60+06:
https://tests.stockfishchess.org/tests/view/664fa166a86388d5e27d7d6b
Tuned on top of: https://github.com/official-stockfish/Stockfish/pull/5287
Passed STC:
https://tests.stockfishchess.org/tests/view/664fc8b7a86388d5e27d8dac
LLR: 2.96 (-2.94,2.94) <0.00,2.00>
Total: 64672 W: 16731 L: 16371 D: 31570
Ptnml(0-2): 239, 7463, 16517, 7933, 184
Passed LTC:
https://tests.stockfishchess.org/tests/view/664fd5f9a86388d5e27d8dfe
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 210648 W: 53489 L: 52813 D: 104346
Ptnml(0-2): 102, 23129, 58164, 23849, 80
closes https://github.com/official-stockfish/Stockfish/pull/5288
Bench: 1717838
2024-05-26 20:32:30 +02:00
Stéphane Nicolet
4d876275cf
Simplify material weights in evaluation
...
This patch uses the same material weights for the nnue
amplification term and the optimism term in evaluate().
STC:
LLR: 2.99 (-2.94,2.94) <-1.75,0.25>
Total: 83360 W: 21489 L: 21313 D: 40558
Ptnml(0-2): 303, 9934, 21056, 10058, 329
https://tests.stockfishchess.org/tests/view/664eee69928b1fb18de500d9
LTC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 192648 W: 48675 L: 48630 D: 95343
Ptnml(0-2): 82, 21484, 53161, 21501, 96
https://tests.stockfishchess.org/tests/view/664fa17aa86388d5e27d7d6e
closes https://github.com/official-stockfish/Stockfish/pull/5287
Bench: 1495602
2024-05-26 20:24:05 +02:00
Linmiao Xu
365aa85dce
Remove material imbalance param when adjusting optimism
...
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/664d033d830eb9f886616aff
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 102144 W: 26283 L: 26135 D: 49726
Ptnml(0-2): 292, 12201, 25991, 12243, 345
Passed non-regression LTC:
https://tests.stockfishchess.org/tests/view/664d5c00830eb9f886616cb3
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 250032 W: 63022 L: 63036 D: 123974
Ptnml(0-2): 103, 27941, 68970, 27871, 131
closes https://github.com/official-stockfish/Stockfish/pull/5284
Bench: 1330940
2024-05-23 21:37:46 +02:00
Linmiao Xu
1dcffa6210
Comment about re-evaluating positions
...
While the smallNet bool is no longer used as of now,
setting it to false upon re-evaluation represents the
correct eval state.
closes https://github.com/official-stockfish/Stockfish/pull/5279
No functional change
2024-05-23 21:30:24 +02:00
Linmiao Xu
c14b69790a
Lower smallnet threshold with updated eval divisors
...
Params found after 30k spsa games at 60+0.6, with initial
values from 64k spsa games at 45+0.45
First spsa with 64k / 120k games at 45+0.45:
https://tests.stockfishchess.org/tests/view/664a561b5fc7b70b8817c663
https://tests.stockfishchess.org/tests/view/664ae88e830eb9f8866146f9
Second spsa with 30k / 120k games at 60+0.6:
https://tests.stockfishchess.org/tests/view/664be227830eb9f886615a36
Values found at 10k games at 60+0.6 also passed STC and LTC:
https://tests.stockfishchess.org/tests/view/664bf4bd830eb9f886615a72
https://tests.stockfishchess.org/tests/view/664c0905830eb9f886615abf
Passed STC:
https://tests.stockfishchess.org/tests/view/664c139e830eb9f886615af2
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 69408 W: 18216 L: 17842 D: 33350
Ptnml(0-2): 257, 8275, 17401, 8379, 392
Passed LTC:
https://tests.stockfishchess.org/tests/view/664cdaf7830eb9f886616a24
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 35466 W: 9075 L: 8758 D: 17633
Ptnml(0-2): 27, 3783, 9794, 4104, 25
closes https://github.com/official-stockfish/Stockfish/pull/5280
bench 1301287
2024-05-21 22:06:17 +02:00
FauziAkram
87bad0c38a
Refine Evaluation Scaling with Piece-Specific Weights
...
Refine Evaluation Scaling with Piece-Specific Weights, instead of the simplified npm method.
I took the initial idea from Viren6 , as he worked on it in September of last year.
I worked on it, and tuned it, and now it passed both tests.
Passed STC:
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 95712 W: 24731 L: 24325 D: 46656
Ptnml(0-2): 363, 11152, 24357, 11684, 300
https://tests.stockfishchess.org/tests/view/664b5493830eb9f886614af3
Passed LTC:
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 204480 W: 52167 L: 51501 D: 100812
Ptnml(0-2): 114, 22579, 56166, 23289, 92
https://tests.stockfishchess.org/tests/view/664b75dd830eb9f886614b44
closes https://github.com/official-stockfish/Stockfish/pull/5277
Bench: 1384337
2024-05-21 08:44:59 +02:00
FauziAkram
b8ccaf038a
Use same shuffling Constant for both nets
...
Passed STC:
https://tests.stockfishchess.org/tests/view/664a42b15fc7b70b8817aeef
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 87840 W: 22759 L: 22594 D: 42487
Ptnml(0-2): 335, 10351, 22324, 10634, 276
Passed LTC:
https://tests.stockfishchess.org/tests/view/664a46995fc7b70b8817af02
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 163122 W: 41443 L: 41367 D: 80312
Ptnml(0-2): 105, 18154, 44927, 18310, 65
closes https://github.com/official-stockfish/Stockfish/pull/5273
bench: 1190174
2024-05-21 08:27:16 +02:00
Linmiao Xu
4d88a63e60
Re-eval only if smallnet output flips from simple eval
...
Recent attempts to change the smallnet nnue re-eval
threshold did not show much elo difference:
https://tests.stockfishchess.org/tests/view/664a29bb25a9058c4d21d53c
https://tests.stockfishchess.org/tests/view/664a299925a9058c4d21d53a
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/664a3ea95fc7b70b8817aee2
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 22304 W: 5905 L: 5664 D: 10735
Ptnml(0-2): 67, 2602, 5603, 2783, 97
Passed non-regression LTC:
https://tests.stockfishchess.org/tests/view/664a43d35fc7b70b8817aef4
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 37536 W: 9667 L: 9460 D: 18409
Ptnml(0-2): 25, 4090, 10321, 4317, 15
closes https://github.com/official-stockfish/Stockfish/pull/5271
bench 1287409
2024-05-21 08:13:25 +02:00
Linmiao Xu
2694fce928
Simplify away adjustEval lambda
...
Now that only the shuffling constant differs between nets,
a lambda for adjusting eval is no longer needed.
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/664806ca6dcff0d1d6b05f34
LLR: 2.99 (-2.94,2.94) <-1.75,0.25>
Total: 31552 W: 8175 L: 7959 D: 15418
Ptnml(0-2): 76, 3180, 9065, 3362, 93
closes https://github.com/official-stockfish/Stockfish/pull/5260
No functional change
2024-05-18 18:08:39 +02:00
Linmiao Xu
99dfc63e03
Use one nnue pawn count multiplier
...
Switch to the value used by the main net.
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/6647e8096dcff0d1d6b05e96
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 51040 W: 13249 L: 13044 D: 24747
Ptnml(0-2): 139, 6029, 13016, 6160, 176
Passed non-regression LTC:
https://tests.stockfishchess.org/tests/view/6647f4a46dcff0d1d6b05eea
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 20460 W: 5195 L: 4972 D: 10293
Ptnml(0-2): 8, 2178, 5637, 2397, 10
https://github.com/official-stockfish/Stockfish/pull/5258
bench 1887462
2024-05-18 09:29:26 +02:00
Linmiao Xu
d92d1f3180
Move smallnet threshold logic into a function
...
Now that the smallnet threshold is no longer a constant,
use a function to organize it with other eval code.
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/66459fa093ce6da3e93b5ba2
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 217600 W: 56281 L: 56260 D: 105059
Ptnml(0-2): 756, 23787, 59729, 23736, 792
closes https://github.com/official-stockfish/Stockfish/pull/5255
No functional change
2024-05-18 09:21:00 +02:00
Linmiao Xu
47597641dc
Lower smallnet threshold linearly as pawn count decreases
...
Passed STC:
https://tests.stockfishchess.org/tests/view/6644f677324e96f42f89d894
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 377920 W: 97135 L: 96322 D: 184463
Ptnml(0-2): 1044, 44259, 97588, 44978, 1091
Passed LTC:
https://tests.stockfishchess.org/tests/view/664548af93ce6da3e93b31b3
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 169056 W: 42901 L: 42312 D: 83843
Ptnml(0-2): 58, 18538, 46753, 19115, 64
closes https://github.com/official-stockfish/Stockfish/pull/5252
Bench: 1991750
2024-05-16 14:19:28 +02:00
Linmiao Xu
541406ab91
Use same nnue divisor for both nets
...
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/6643ceeabc537f56194506f6
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 224800 W: 57910 L: 57896 D: 108994
Ptnml(0-2): 673, 26790, 57519, 26686, 732
Passed non-regression LTC:
https://tests.stockfishchess.org/tests/view/6643ff15bc537f5619451719
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 347658 W: 87574 L: 87688 D: 172396
Ptnml(0-2): 207, 39004, 95488, 38956, 174
closes https://github.com/official-stockfish/Stockfish/pull/5250
Bench: 1804704
2024-05-16 14:16:54 +02:00
Linmiao Xu
1f3a0fda2e
Use same eval divisor for both nets
...
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/66428f146577e9d2c8a29cf8
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 241024 W: 62173 L: 62177 D: 116674
Ptnml(0-2): 904, 28648, 61407, 28654, 899
Passed non-regression LTC:
https://tests.stockfishchess.org/tests/view/6643ae6f1f32a966da74977b
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 193710 W: 48762 L: 48717 D: 96231
Ptnml(0-2): 70, 21599, 53481, 21626, 79
closes https://github.com/official-stockfish/Stockfish/pull/5246
Bench: 1700680
2024-05-15 16:26:00 +02:00
mstembera
09dba1f080
Call adjustEval with correct parameters after rescore
...
Set smallNet to false after rescoring so we call adjustEval() w/ correct
parameters.
STC:
https://tests.stockfishchess.org/tests/view/664308687134c82f3f7a4003
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 146912 W: 37856 L: 37756 D: 71300
Ptnml(0-2): 566, 17562, 37122, 17618, 588
LTC:
https://tests.stockfishchess.org/tests/view/6643a0821f32a966da7485d6
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 390414 W: 98015 L: 98173 D: 194226
Ptnml(0-2): 162, 43555, 107929, 43401, 160
closes https://github.com/official-stockfish/Stockfish/pull/5244
Bench: 1819318
2024-05-15 16:19:04 +02:00
Linmiao Xu
0b08953174
Re-evaluate some small net positions for more accurate evals
...
Use main net evals when small net evals hint that higher eval
accuracy may be worth the slower eval speeds. With Finny caches,
re-evals with the main net are less expensive than before.
Original idea by mstembera who I've added as co-author to this PR.
Based on reEval tests by mstembera:
https://tests.stockfishchess.org/tests/view/65e69187b6345c1b934866e5
https://tests.stockfishchess.org/tests/view/65e863aa0ec64f0526c3e991
A few variants of this patch also passed LTC:
https://tests.stockfishchess.org/tests/view/663d2108507ebe1c0e91f407
https://tests.stockfishchess.org/tests/view/663e388c3a2f9702074bc152
Passed STC:
https://tests.stockfishchess.org/tests/view/663dadbd1a61d6377f190e2c
LLR: 2.93 (-2.94,2.94) <0.00,2.00>
Total: 92320 W: 23941 L: 23531 D: 44848
Ptnml(0-2): 430, 10993, 22931, 11349, 457
Passed LTC:
https://tests.stockfishchess.org/tests/view/663ef48b2948bf9aa698690c
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 98934 W: 24907 L: 24457 D: 49570
Ptnml(0-2): 48, 10952, 27027, 11382, 58
closes https://github.com/official-stockfish/Stockfish/pull/5238
bench 1876282
Co-Authored-By: mstembera <5421953+mstembera@users.noreply.github.com>
2024-05-13 07:30:18 +02:00
Linmiao Xu
53f363041c
Simplify npm constants when adjusting eval
...
Passed non-regression STC:
https://tests.stockfishchess.org/tests/view/663d0c4f507ebe1c0e91ec8d
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 162784 W: 41987 L: 41906 D: 78891
Ptnml(0-2): 520, 19338, 41591, 19427, 516
Passed non-regression LTC:
https://tests.stockfishchess.org/tests/view/663d20fd507ebe1c0e91f405
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 457242 W: 115022 L: 115250 D: 226970
Ptnml(0-2): 271, 51566, 125179, 51330, 275
closes https://github.com/official-stockfish/Stockfish/pull/5237
Bench: 2238216
2024-05-13 07:25:22 +02:00
FauziAkram
d712ed38d1
Simplify shuffling and optimism divisors to constants
...
Shuffling divisor and Optimism divisors passed STC & LTC separately:
shuf STC: https://tests.stockfishchess.org/tests/view/66356316b4e9bdbc7228b995
shuf LTC: https://tests.stockfishchess.org/tests/view/6635815a73559a8aa857c1dc
opt STC: https://tests.stockfishchess.org/tests/view/66356326b4e9bdbc7228b9a0
opt LTC: https://tests.stockfishchess.org/tests/view/663615c673559a8aa8589f8a
And then passed LTC together:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 178278 W: 45039 L: 44979 D: 88260
Ptnml(0-2): 43, 19776, 49460, 19798, 62
https://tests.stockfishchess.org/tests/view/66363f19cdb7cf5da64e22a3
closes https://github.com/official-stockfish/Stockfish/pull/5212
Bench: 2198243
2024-05-05 13:02:42 +02:00
cj5716
8ee9905d8b
Remove PSQT-only mode
...
Passed STC:
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 94208 W: 24270 L: 24112 D: 45826
Ptnml(0-2): 286, 11186, 24009, 11330, 293
https://tests.stockfishchess.org/tests/view/6635ddd773559a8aa8582826
Passed LTC:
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 114960 W: 29107 L: 28982 D: 56871
Ptnml(0-2): 37, 12683, 31924, 12790, 46
https://tests.stockfishchess.org/tests/view/663604a973559a8aa85881ed
closes #5214
Bench 1653939
2024-05-05 12:36:20 +02:00
Ciekce
eb20de36c0
Avoid unnecessary creation of accumulator cache
...
Saves a (currently) 800 KB allocation and deallocation when running
`eval`, not particularly significant and zero impact on play but not
necessary either.
closes https://github.com/official-stockfish/Stockfish/pull/5201
No functional change
2024-05-01 14:10:57 +02:00
Stefan Geschwentner
48a3b7c0ee
Simplify non-pawn material divisor to a constant
...
Passed STC:
https://tests.stockfishchess.org/tests/view/662942603fe04ce4cefc7aba
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 272832 W: 70456 L: 70497 D: 131879
Ptnml(0-2): 1020, 32619, 69154, 32628, 995
Passed LTC:
https://tests.stockfishchess.org/tests/view/662dfe3b6115ff6764c829eb
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 100254 W: 25446 L: 25303 D: 49505
Ptnml(0-2): 121, 11292, 27166, 11419, 129
closes https://github.com/official-stockfish/Stockfish/pull/5198
Bench: 1544645
2024-04-28 21:43:46 +02:00
mstembera
a129c0695b
Combine remove and add in update_accumulator_refresh_cache()
...
Combine remove and add in update_accumulator_refresh_cache().
Move remove before add to match other parts of the code.
STC:
https://tests.stockfishchess.org/tests/view/662d96dc6115ff6764c7f4ca
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 364032 W: 94421 L: 93624 D: 175987
Ptnml(0-2): 1261, 41983, 94811, 42620, 1341
closes https://github.com/official-stockfish/Stockfish/pull/5194
Bench: 1836777
2024-04-28 21:35:48 +02:00
mstembera
940a3a7383
Cache small net w/ psqtOnly support
...
Caching the small net in the same way as the big net allows them to
share the same code path and completely removes
update_accumulator_refresh().
STC:
https://tests.stockfishchess.org/tests/view/662bfb5ed46f72253dcfed85
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 151712 W: 39252 L: 39158 D: 73302
Ptnml(0-2): 565, 17474, 39683, 17570, 564
closes https://github.com/official-stockfish/Stockfish/pull/5194
Bench: 1836777
2024-04-28 21:30:19 +02:00
Disservin
3502c8ae42
Fix missing initialization of AccumulatorCaches in Eval::trace
...
Add a constructor to `AccumulatorCaches` instead of just calling
`clear(networks)` to prevent similar issues from appearing in the
future.
fixes https://github.com/official-stockfish/Stockfish/issues/5190
closes https://github.com/official-stockfish/Stockfish/pull/5191
No functional change
2024-04-28 21:26:36 +02:00
gab8192
49ef4c935a
Implement accumulator refresh table
...
For each thread persist an accumulator cache for the network, where each
cache contains multiple entries for each of the possible king squares.
When the accumulator needs to be refreshed, the cached entry is used to more
efficiently update the accumulator, instead of rebuilding it from scratch.
This idea, was first described by Luecx (author of Koivisto) and
is commonly referred to as "Finny Tables".
When the accumulator needs to be refreshed, instead of filling it with
biases and adding every piece from scratch, we...
1. Take the `AccumulatorRefreshEntry` associated with the new king bucket
2. Calculate the features to activate and deactivate (from differences
between bitboards in the entry and bitboards of the actual position)
3. Apply the updates on the refresh entry
4. Copy the content of the refresh entry accumulator to the accumulator
we were refreshing
5. Copy the bitboards from the position to the refresh entry, to match
the newly updated accumulator
Results at STC:
https://tests.stockfishchess.org/tests/view/662301573fe04ce4cefc1386
(first version)
https://tests.stockfishchess.org/tests/view/6627fa063fe04ce4cefc6560
(final)
Non-Regression between first and final:
https://tests.stockfishchess.org/tests/view/662801e33fe04ce4cefc660a
STC SMP:
https://tests.stockfishchess.org/tests/view/662808133fe04ce4cefc667c
closes https://github.com/official-stockfish/Stockfish/pull/5183
No functional change
2024-04-24 18:38:20 +02:00
FauziAkram
fcba524793
Tune Search Parameters
...
Parameters Tune, adding also another tunable parameter (npmDiv) to be
variable for different nets (bignet, smallnet, psqtOnly smallnet). P.s:
The changed values are only the parameters where there is agreement
among the different time controls, so in other words, the tunings are
telling us that changing these specific values to this specific
direction is good in all time controls, so there shouldn't be a high
risk of regressing at longer time controls.
Passed STC:
LLR: 2.97 (-2.94,2.94) <0.00,2.00>
Total: 39552 W: 10329 L: 9999 D: 19224
Ptnml(0-2): 156, 4592, 9989, 4844, 195
https://tests.stockfishchess.org/tests/view/661be9a0bd68065432a088c0
Passed LTC:
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 56394 W: 14439 L: 14078 D: 27877
Ptnml(0-2): 30, 6152, 15480, 6497, 38
https://tests.stockfishchess.org/tests/view/661c746296961e72eb565406
closes https://github.com/official-stockfish/Stockfish/pull/5187
Bench: 1836777
2024-04-24 18:24:01 +02:00
Disservin
299707d2c2
Split UCI into UCIEngine and Engine
...
This is another refactor which aims to decouple uci from stockfish. A new engine
class manages all engine related logic and uci is a "small" wrapper around it.
In the future we should also try to remove the need for the Position object in
the uci and replace the options with an actual options struct instead of using a
map. Also convert the std::string's in the Info structs a string_view.
closes #5147
No functional change
2024-04-04 00:15:17 +02:00
Gahtan Nahdi
ed24e3a0a6
Remove material imbalance from nnue eval
...
Passed non-reg STC:
https://tests.stockfishchess.org/tests/view/65fdf11f0ec64f0526c52b57
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 76480 W: 19893 L: 19712 D: 36875
Ptnml(0-2): 339, 9107, 19157, 9308, 329
Passed non-reg LTC:
https://tests.stockfishchess.org/tests/view/65fee22e0ec64f0526c53885
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 150948 W: 38078 L: 37988 D: 74882
Ptnml(0-2): 111, 16997, 41148, 17127, 91
closes https://github.com/official-stockfish/Stockfish/pull/5135
Bench: 2103324
2024-03-26 18:17:00 +01:00
Gahtan Nahdi
1a6c22c511
Evaluation adjustment for different eval types
...
Gives different eval scaling parameters for the three different types
of evaluation (bignet, smallnet, psqtOnly).
Passed STC:
https://tests.stockfishchess.org/tests/view/65f4b0020ec64f0526c4a3bd
LLR: 2.96 (-2.94,2.94) <0.00,2.00>
Total: 168064 W: 43507 L: 42987 D: 81570
Ptnml(0-2): 662, 19871, 42445, 20393, 661
Passed LTC:
https://tests.stockfishchess.org/tests/view/65f6be1a0ec64f0526c4c361
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 162564 W: 41188 L: 40604 D: 80772
Ptnml(0-2): 120, 18112, 44216, 18732, 102
closes https://github.com/official-stockfish/Stockfish/pull/5122
Bench: 2113576
2024-03-20 16:29:35 +01:00
Robert Nurnberg
9b92ada935
Base WDL model on material count and normalize evals dynamically
...
This PR proposes to change the parameter dependence of Stockfish's
internal WDL model from full move counter to material count. In addition
it ensures that an evaluation of 100 centipawns always corresponds to a
50% win probability at fishtest LTC, whereas for master this holds only
at move number 32. See also
https://github.com/official-stockfish/Stockfish/pull/4920 and the
discussion therein.
The new model was fitted based on about 340M positions extracted from
5.6M fishtest LTC games from the last three weeks, involving SF versions
from e67cc979fdhttps://github.com/official-stockfish/WDL_model ) are:
```
./updateWDL.sh --firstrev e67cc979fdhttps://github.com/official-stockfish/Stockfish/pull/5121
No functional change
2024-03-20 16:29:35 +01:00
FauziAkram
627974c99f
Search + Eval + Movepick Tune
...
Passed STC:
https://tests.stockfishchess.org/tests/view/65ef15220ec64f0526c44b04
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 24480 W: 6459 L: 6153 D: 11868
Ptnml(0-2): 101, 2798, 6184, 3008, 149
Passed LTC:
https://tests.stockfishchess.org/tests/view/65ef4bac0ec64f0526c44f50
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 53316 W: 13561 L: 13203 D: 26552
Ptnml(0-2): 27, 5925, 14408, 6259, 39
closes https://github.com/official-stockfish/Stockfish/pull/5104
Bench: 1715522
2024-03-12 16:47:11 +01:00
Disservin
1a26d698de
Refactor Network Usage
...
Continuing from PR #4968 , this update improves how Stockfish handles network
usage, making it easier to manage and modify networks in the future.
With the introduction of a dedicated Network class, creating networks has become
straightforward. See uci.cpp:
```cpp
NN::NetworkBig({EvalFileDefaultNameBig, "None", ""}, NN::embeddedNNUEBig)
```
The new `Network` encapsulates all network-related logic, significantly reducing
the complexity previously required to support multiple network types, such as
the distinction between small and big networks #4915 .
Non-Regression STC:
https://tests.stockfishchess.org/tests/view/65edd26c0ec64f0526c43584
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 33760 W: 8887 L: 8661 D: 16212
Ptnml(0-2): 143, 3795, 8808, 3961, 173
Non-Regression SMP STC:
https://tests.stockfishchess.org/tests/view/65ed71970ec64f0526c42fdd
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 59088 W: 15121 L: 14931 D: 29036
Ptnml(0-2): 110, 6640, 15829, 6880, 85
Compiled with `make -j profile-build`
```
bash ./bench_parallel.sh ./stockfish ./stockfish-nnue 13 50
sf_base = 1568540 +/- 7637 (95%)
sf_test = 1573129 +/- 7301 (95%)
diff = 4589 +/- 8720 (95%)
speedup = 0.29260% +/- 0.556% (95%)
```
Compiled with `make -j build`
```
bash ./bench_parallel.sh ./stockfish ./stockfish-nnue 13 50
sf_base = 1472653 +/- 7293 (95%)
sf_test = 1491928 +/- 7661 (95%)
diff = 19275 +/- 7154 (95%)
speedup = 1.30886% +/- 0.486% (95%)
```
closes https://github.com/official-stockfish/Stockfish/pull/5100
No functional change
2024-03-12 16:41:08 +01:00
Muzhen Gaming
10e2732978
VVLTC search tune
...
Result of 32k games of tuning at 60+0.6 8-thread. Link to the tuning
attempt:
https://tests.stockfishchess.org/tests/view/65def7b04b19edc854ebdec8
Passed VVLTC first SPRT:
https://tests.stockfishchess.org/tests/view/65e51b53416ecd92c162ab7f
LLR: 2.94 (-2.94,2.94) <0.00,2.00>
Total: 37570 W: 9613 L: 9342 D: 18615
Ptnml(0-2): 2, 3454, 11601, 3727, 1
Passed VVLTC second SPRT:
https://tests.stockfishchess.org/tests/view/65e87d1c0ec64f0526c3eb39
LLR: 2.94 (-2.94,2.94) <0.50,2.50>
Total: 123158 W: 31463 L: 31006 D: 60689
Ptnml(0-2): 5, 11589, 37935, 12044, 6
Note: The small net and psqt-only thresholds have been moved to
evaluate.h. The reasoning is that these values are used in both
`evaluate.cpp` and `evaluate_nnue.cpp`, and thus unifying their usage
avoids inconsistencies during testing, where one occurrence is changed
without the other (this happened during the search tune SPRT).
closes https://github.com/official-stockfish/Stockfish/pull/5101
Bench: 1741218
2024-03-11 10:04:37 +01:00
mstembera
7831131591
Only evaluate the PSQT part of the small net for large evals.
...
Thanks to Viren6 for suggesting to set complexity to 0.
STC https://tests.stockfishchess.org/tests/view/65d7d6709b2da0226a5a203f
LLR: 2.92 (-2.94,2.94) <0.00,2.00>
Total: 328384 W: 85316 L: 84554 D: 158514
Ptnml(0-2): 1414, 39076, 82486, 39766, 1450
LTC https://tests.stockfishchess.org/tests/view/65dce6d290f639b028a54d2e
LLR: 2.95 (-2.94,2.94) <0.50,2.50>
Total: 165162 W: 41918 L: 41330 D: 81914
Ptnml(0-2): 102, 18332, 45124, 18922, 101
closes https://github.com/official-stockfish/Stockfish/pull/5083
bench: 1504003
2024-03-03 15:29:58 +01:00
Disservin
d07033d5da
Expose EvalFileSmall option for small net
...
Since https://github.com/official-stockfish/fishtest/pull/1870 has been merged
it's time for this update.
5k Fixed Games showed no problems.
https://tests.stockfishchess.org/tests/view/65d9cc274c0e22b904f574d7
closes https://github.com/official-stockfish/Stockfish/pull/5068
No functional change
2024-02-24 17:57:49 +01:00
Disservin
9068fdc57b
Assorted cleanups
...
Assorted cleanups
closes https://github.com/official-stockfish/Stockfish/pull/5046
No functional change
Co-Authored-By: Shahin M. Shahin <41402573+peregrineshahin@users.noreply.github.com>
Co-Authored-By: cj5716 <125858804+cj5716@users.noreply.github.com>
2024-02-11 19:52:00 +01:00
cj5716
c0107b3c27
Remove simple eval
...
With the recent introduction of the dual NNUE, the
need for simple eval is no longer there.
Passed STC:
https://tests.stockfishchess.org/tests/view/65c1f735c865510db0281652
LLR: 2.96 (-2.94,2.94) <-1.75,0.25>
Total: 85312 W: 22009 L: 21837 D: 41466
Ptnml(0-2): 334, 10155, 21567, 10205, 395
Passed LTC:
https://tests.stockfishchess.org/tests/view/65c2d64bc865510db0282810
LLR: 2.95 (-2.94,2.94) <-1.75,0.25>
Total: 49956 W: 12596 L: 12402 D: 24958
Ptnml(0-2): 28, 5553, 13624, 5743, 30
closes https://github.com/official-stockfish/Stockfish/pull/5037
Bench 1213676
2024-02-09 19:06:25 +01:00
Disservin
88331add0d
Remove the dependency on a Worker from evaluate
...
Also remove dead code, `rootSimpleEval` is no longer used since the introduction of dual net.
`iterBestValue` is also no longer used in evaluate and can be reduced to a local variable.
closes https://github.com/official-stockfish/Stockfish/pull/4979
No functional change
2024-01-14 10:46:13 +01:00
Disservin
a107910951
Refactor global variables
...
This aims to remove some of the annoying global structure which Stockfish has.
Overall there is no major elo regression to be expected.
Non regression SMP STC (paused, early version):
https://tests.stockfishchess.org/tests/view/65983d7979aa8af82b9608f1
LLR: 0.23 (-2.94,2.94) <-1.75,0.25>
Total: 76232 W: 19035 L: 19096 D: 38101
Ptnml(0-2): 92, 8735, 20515, 8690, 84
Non regression STC (early version):
https://tests.stockfishchess.org/tests/view/6595b3a479aa8af82b95da7f
LLR: 2.93 (-2.94,2.94) <-1.75,0.25>
Total: 185344 W: 47027 L: 46972 D: 91345
Ptnml(0-2): 571, 21285, 48943, 21264, 609
Non regression SMP STC:
https://tests.stockfishchess.org/tests/view/65a0715c79aa8af82b96b7e4
LLR: 2.94 (-2.94,2.94) <-1.75,0.25>
Total: 142936 W: 35761 L: 35662 D: 71513
Ptnml(0-2): 209, 16400, 38135, 16531, 193
These global structures/variables add hidden dependencies and allow data
to be mutable from where it shouldn't it be (i.e. options). They also
prevent Stockfish from internal selfplay, which would be a nice thing to
be able to do, i.e. instantiate two Stockfish instances and let them
play against each other. It will also allow us to make Stockfish a
library, which can be easier used on other platforms.
For consistency with the old search code, `thisThread` has been kept,
even though it is not strictly necessary anymore. This the first major
refactor of this kind (in recent time), and future changes are required,
to achieve the previously described goals. This includes cleaning up the
dependencies, transforming the network to be self contained and coming
up with a plan to deal with proper tablebase memory management (see
comments for more information on this).
The removal of these global structures has been discussed in parts with
Vondele and Sopel.
closes https://github.com/official-stockfish/Stockfish/pull/4968
No functional change
2024-01-13 19:40:53 +01:00
Disservin
99cdb920fc
Cleanup Evalfile handling
...
This cleans up the EvalFile handling after the merge of #4915 ,
which has become a bit confusing on what it is actually doing.
closes https://github.com/official-stockfish/Stockfish/pull/4971
No functional change
2024-01-08 18:33:38 +01:00
