SolsticeBench

Finished
TestTCEngineAuthorResult
B83DA96E79627EBE SPRT diff
lmr-statscore (features #42): LMR history term 2*main + cont
4.0+0.04SolsticeDragonLLR: -1.57 (-2.94, 2.94) [0.00, 5.00]
Games: 2048 W: 458 L: 509 D: 1081
Ptnml(0-2): 13, 275, 496, 230, 10
CAE25101556AEE98 SPRT diff
se-ttpv (features #59): singular margin + 43/16 per ply at tt-pv non-PV nodes
4.0+0.04SolsticeDragonLLR: -0.13 (-2.94, 2.94) [0.00, 5.00]
Games: 7936 W: 1873 L: 1847 D: 4216
Ptnml(0-2): 59, 928, 1966, 958, 57
5B226C62C315DE36 SPRT diff
rfp-ttquiet (features #20): RFP only without a quiet TT move
4.0+0.04SolsticeDragonLLR: -0.21 (-2.94, 2.94) [0.00, 5.00]
Games: 2048 W: 485 L: 485 D: 1078
Ptnml(0-2): 19, 242, 493, 260, 10
1D11D5582B039F33 SPRT diff
nmp-imp (features #22): NMP needs eval >= beta + 48 raw when not improving
4.0+0.04SolsticeDragonLLR: -0.70 (-2.94, 2.94) [0.00, 5.00]
Games: 2048 W: 461 L: 479 D: 1108
Ptnml(0-2): 13, 251, 512, 237, 11
B7836A4DB83FD496 SPRT diff
tt-bucket2 (features #73): 2-entry TT buckets, replace lowest depth - 8*age
4.0+0.04SolsticeDragonLLR: -0.72 (-2.94, 2.94) [0.00, 5.00]
Games: 2048 W: 452 L: 471 D: 1125
Ptnml(0-2): 15, 248, 515, 233, 13
0BA0B673B3B82F51 SPRT diff
hist-linear (features #7): history bonus/malus linear in depth (300d-250 / 230d-190)
4.0+0.04SolsticeDragonLLR: 2.97 (-2.94, 2.94) [0.00, 5.00]
Games: 8218 W: 2017 L: 1869 D: 4332
Ptnml(0-2): 52, 962, 1971, 1034, 90
D7A84D3630333B32 SPRT diff
lmr-quietfrac (features #47): quiet LMR index scaled by the number of quiets, as lmp-frac
4.0+0.04SolsticeDragonLLR: -0.64 (-2.94, 2.94) [0.00, 5.00]
Games: 2048 W: 482 L: 499 D: 1067
Ptnml(0-2): 19, 267, 473, 242, 23
5BF53E6B65F5FA4D SPRT diff
cont-ply4 (features #4): 4-ply continuation history, half weight in ordering/LMR
4.0+0.04SolsticeDragonLLR: -0.44 (-2.94, 2.94) [0.00, 5.00]
Games: 2048 W: 488 L: 497 D: 1063
Ptnml(0-2): 24, 242, 493, 249, 16
37E5378B35A54AD3 SPRT diff
lmr-deeper2 (features #45): deeper/shallower LMR re-search, margins 70+2d / 11
4.0+0.04SolsticeDragonLLR: -0.19 (-2.94, 2.94) [0.00, 5.00]
Games: 2048 W: 475 L: 474 D: 1099
Ptnml(0-2): 17, 256, 481, 249, 21
93D3C0C60FAEDDA2 SPRT diff
qs-checks (features #66): quiet direct checks (SEE>=0) at the first quiescence ply
4.0+0.04SolsticeDragonLLR: -0.52 (-2.94, 2.94) [0.00, 5.00]
Games: 1024 W: 226 L: 242 D: 556
Ptnml(0-2): 11, 129, 243, 123, 6
DE8FC3E63A2B6029 SPRT diff
ttpv-graded (features #44): tt-pv LMR + less when tt score > alpha / tt depth >= depth
4.0+0.04SolsticeDragonLLR: -0.53 (-2.94, 2.94) [0.00, 5.00]
Games: 512 W: 108 L: 126 D: 278
Ptnml(0-2): 6, 66, 126, 56, 2
DE777F12092CF19B SPRT diff
depth-floor1 (features #55): LMR reduced depth >= 1, not 0
4.0+0.04SolsticeDragonLLR: -0.63 (-2.94, 2.94) [0.00, 5.00]
Games: 768 W: 164 L: 185 D: 419
Ptnml(0-2): 6, 106, 178, 91, 3
A467AED3DFF3B05D SPRT diff
lmr-depth-prune2 (features #34): quiet futility/SEE pruning by the LMR depth
4.0+0.04SolsticeDragonLLR: -1.06 (-2.94, 2.94) [0.00, 5.00]
Games: 578 W: 110 L: 149 D: 319
Ptnml(0-2): 10, 83, 133, 62, 1
4C15DE7B5A30C5DB SPRT diff
wdl-margins (idea #37): RFP/futility/razor margins x a(material)/a(58) from the refitted win-rate model
4.0+0.04SolsticeDragonLLR: -1.37 (-2.94, 2.94) [0.00, 5.00]
Games: 1792 W: 394 L: 439 D: 959
Ptnml(0-2): 14, 236, 440, 193, 13
A742FC5FD50429DA SPRT diff
see-hist (idea #72): quiet SEE pruning margin scaled by quiet history (32/16384, capped at 0)
4.0+0.04SolsticeDragonLLR: 0.19 (-2.94, 2.94) [0.00, 5.00]
Games: 9728 W: 2314 L: 2268 D: 5146
Ptnml(0-2): 61, 1182, 2358, 1176, 87
7BD7C88F92F2688B SPRT diff
threat-hist2 (features #13): butterfly history [from attacked][to attacked], port of #65
4.0+0.04SolsticeDragonLLR: 3.00 (-2.94, 2.94) [0.00, 5.00]
Games: 2714 W: 688 L: 561 D: 1465
Ptnml(0-2): 15, 283, 649, 380, 30
D5CB60941E3ABDD3 SPRT diff
escape-lmr (idea #74): LMR one ply less for quiet moves of pieces attacked by an enemy pawn
4.0+0.04SolsticeDragonLLR: -0.12 (-2.94, 2.94) [0.00, 5.00]
Games: 5376 W: 1261 L: 1244 D: 2871
Ptnml(0-2): 47, 644, 1290, 659, 48
A7D09541C8EAE7E8 SPRT diff
node-hist (idea #53): quiet history bonus scaled by avg cutoff subtree / this cutoff's subtree, clamp [0.5, 2]
4.0+0.04SolsticeDragonLLR: -0.67 (-2.94, 2.94) [0.00, 5.00]
Games: 2048 W: 461 L: 479 D: 1108
Ptnml(0-2): 16, 271, 468, 253, 16
3DC78F15BFD6D794 SPRT diff
rfp-tension (idea #3): RFP margin + (tension-100)/2 raw where tension = eval(stm)+eval(other stm)
4.0+0.04SolsticeDragonLLR: -1.27 (-2.94, 2.94) [0.00, 5.00]
Games: 1024 W: 224 L: 269 D: 531
Ptnml(0-2): 15, 130, 258, 103, 6
11754C243D863D23 SPRT diff
tt-surprise (idea #25): LMR one ply less where |TT score - TT raw eval| > 200 (depth>=4 lower/exact entries)
4.0+0.04SolsticeDragonLLR: -0.68 (-2.94, 2.94) [0.00, 5.00]
Games: 2560 W: 593 L: 609 D: 1358
Ptnml(0-2): 24, 304, 635, 298, 19
53DF3C786A6502A5 SPRT diff
corr-order (idea #15): correction lookahead in move ordering. Laptop: -3.2 +-13.1 @1.3k
4.0+0.04SolsticeDragonLLR: -0.92 (-2.94, 2.94) [0.00, 5.00]
Games: 3584 W: 829 L: 850 D: 1905
Ptnml(0-2): 31, 435, 873, 430, 23
3802AEE79C830F59 SPRT diff
corr-spread (idea #14): LMR from correction-table disagreement. Laptop: -4.4 +-8.9 @2.9k
4.0+0.04SolsticeDragonLLR: -0.30 (-2.94, 2.94) [0.00, 5.00]
Games: 8436 W: 1956 L: 1934 D: 4546
Ptnml(0-2): 65, 1021, 2036, 1019, 77
13115DABB980C507 SPRT diff
lmp-frac (idea #62): LMP by the fraction of quiets searched. Laptop: +10.7 +-13.2 @1.3k
4.0+0.04SolsticeDragonLLR: 2.97 (-2.94, 2.94) [0.00, 5.00]
Games: 4848 W: 1204 L: 1071 D: 2573
Ptnml(0-2): 35, 522, 1183, 643, 41
DE688A88D11107BA SPRT diff
net: g456_kb4ob8_h512_w025_sb400 (kb4 + 8 output buckets, h512 gen456 400 SB; gen6 val 0.46075 vs master 0.46098)
4.0+0.04SolsticeDragonLLR: -2.99 (-2.94, 2.94) [0.00, 5.00]
Games: 2036 W: 456 L: 568 D: 1012
Ptnml(0-2): 32, 283, 486, 199, 18
D942D8904D89F864 SPRT diff
net: g456_kb4_h1024_w025_sb400 (kb4 h1024 gen456, 400 SB; gen6 val 0.45939 vs master 0.46098; ~23% fewer nps)
4.0+0.04SolsticeDragonLLR: -2.97 (-2.94, 2.94) [0.00, 5.00]
Games: 4138 W: 982 L: 1082 D: 2074
Ptnml(0-2): 47, 539, 984, 465, 34