Stockfish Testing Queue

Cores

1991

Nodes / sec

1293M

Games / min

1146

Time remaining

52.5h

Workers - 154 machines

Pending approval - 0 tests

No tests pending approval

Paused - 12 tests

26-01-04 sg
tweak_lmr_scale_a2 diff 
  
  
    LLR: -1.39 (-2.94,2.94) <0.00,2.00>
      

    Total: 152384 W: 39065 L: 38964 D: 74355
      

    Ptnml(0-2): 448, 18023, 39152, 18118, 451
sprt @ 10+0.1 th 1
cores: 0 (0)
Use following depth dependent reduction factor: r * depth / (depth * depth + 2).
26-01-04 sg
tweak_lmr_scale_a4 diff 
  
  
    LLR: -1.31 (-2.94,2.94) <0.00,2.00>
      

    Total: 162272 W: 41781 L: 41658 D: 78833
      

    Ptnml(0-2): 520, 19120, 41725, 19259, 512
sprt @ 10+0.1 th 1
cores: 0 (0)
Use following depth dependent reduction factor: r * depth / (depth * depth + 4).
26-01-04 sg
tweak_lmr_scale_a5 diff 
  
  
    LLR: -0.53 (-2.94,2.94) <0.00,2.00>
      

    Total: 90720 W: 23338 L: 23252 D: 44130
      

    Ptnml(0-2): 233, 10739, 23369, 10747, 272
sprt @ 10+0.1 th 1
cores: 0 (0)
Use following depth dependent reduction factor: r * (depth + 1) / (depth * depth + 3).
26-01-04 sg
tweak_lmr_scale_a6 diff 
  
  
    LLR: -1.97 (-2.94,2.94) <0.00,2.00>
      

    Total: 131072 W: 33725 L: 33707 D: 63640
      

    Ptnml(0-2): 411, 15433, 33759, 15593, 340
sprt @ 10+0.1 th 1
cores: 0 (0)
Use following depth dependent reduction factor: r * depth / (depth * depth + 5).
26-01-15 sg
tt_stm3 diff 
  
  
    LLR: -1.19 (-2.94,2.94) <-1.75,0.25>
      

    Total: 22784 W: 5861 L: 5991 D: 10932
      

    Ptnml(0-2): 72, 2717, 5934, 2607, 62
sprt @ 10+0.1 th 1
cores: 0 (0)
Test first as non-regression for default hash setting. Try an alternative and probably faster implementation of hash collision problem with opposite positions. The side to move information is now maintained in the lowest bit in the position key by modifing the zobrist keys directly. The side key has as lowest bit a 1 and for all other keys the lowest bit is set zero (all other zobrist key bits are the same as master)- Also to avoid a endless loop during generation of the cuckoo table the both hash function have be changed not to use the lowest bit because it will now everytime 1. In the pawn_key function then only the rule50 adjusting have to be shift 1 bit up to not effect the lowest bit of the key.
26-01-16 sg
tuned_lmr_conditions2c diff 
  
  
    LLR: -2.02 (-2.94,2.94) <0.00,2.00>
      

    Total: 24128 W: 6148 L: 6291 D: 11689
      

    Ptnml(0-2): 61, 2910, 6281, 2735, 77
sprt @ 10+0.1 th 1
cores: 0 (0)
Double effect: After 92k LTC lmr conditions tuning test most significant negative param A[0][1][2] but use opposite precondition ttPv (the original precondition failed and allowing both struggles but seem at least neutral).
26-01-14 Dan
cm_mc_b_2 diff 
  
  
    LLR: -2.05 (-2.94,2.94) <0.00,2.00>
      

    Total: 150528 W: 38784 L: 38745 D: 72999
      

    Ptnml(0-2): 459, 17768, 38794, 17761, 482
sprt @ 10+0.1 th 1
cores: 0 (0)
take 2
26-01-15 sni
style^ diff 
  
  
    LLR: -1.14 (-2.94,2.94) <0.00,2.00>
      

    Total: 68000 W: 17431 L: 17432 D: 33137
      

    Ptnml(0-2): 193, 8170, 17315, 8089, 233
sprt @ 10+0.1 th 1
cores: 0 (0)
Kortchnoi style = 3 when simple*nnue < 0
26-01-16 sni
style diff 
  
  
    LLR: -1.07 (-2.94,2.94) <0.00,2.00>
      

    Total: 32768 W: 8340 L: 8387 D: 16041
      

    Ptnml(0-2): 127, 3869, 8407, 3886, 95
sprt @ 10+0.1 th 1
cores: 0 (0)
Korchnoi when simple*nnue<0, Tal when simple*nnue>0
26-01-05 0x5
exp-gbquietsort diff 
  
  
    LLR: -1.71 (-2.94,2.94) <-1.75,0.25>
      

    Total: 80064 W: 20503 L: 20744 D: 38817
      

    Ptnml(0-2): 244, 9679, 20461, 9370, 278
sprt @ 10+0.1 th 1
cores: 0 (0)
don't sort bad quiets fixed
26-01-05 0x5
exp-gbquietsort diff 
  
  
    LLR: 2.37 (-2.94,2.94) <-1.75,0.25>
      

    Total: 128480 W: 33207 L: 33138 D: 62135
      

    Ptnml(0-2): 421, 15097, 33133, 15170, 419
sprt @ 10+0.1 th 1
cores: 0 (0)
consider anything we sort as good
26-01-13 pb0
retro diff 
  
  
    LLR: 0.08 (-2.94,2.94) <-1.75,0.25>
      

    Total: 22032 W: 5551 L: 5567 D: 10914
      

    Ptnml(0-2): 9, 2421, 6176, 2397, 13
sprt @ 60+0.6 th 1
cores: 0 (0)
LTC: Take 2: version 2 of a robust retrograde analysis feature. More clean implemented than take 1 and less invasive for the main search. Curious the see if this regresses at normal game play. P.S.: Stopped because just detected it obeys no more nodes limit (so bugged version)

Failed - 0 tests

No failed tests on this page

Active - 34 tests

26-01-14 jay
imp2 diff 
  
  
    LLR: 1.89 (-2.94,2.94) <0.00,2.00>
      

    Total: 518944 W: 134265 L: 133333 D: 251346
      

    Ptnml(0-2): 1568, 61126, 133203, 61956, 1619
sprt @ 10+0.1 th 1
cores: 112 (7)
qsearch: lower futilityBase when improving
26-01-18 fau
antid2 diff 
  
  
    LLR: 1.28 (-2.94,2.94) <0.00,2.00>
      

    Total: 13440 W: 3515 L: 3382 D: 6543
      

    Ptnml(0-2): 32, 1541, 3456, 1644, 47
sprt @ 10+0.1 th 1
cores: 22 (3)
antid2
26-01-18 kev
probcut-reduction diff 
  
  
    LLR: 1.01 (-2.94,2.94) <0.00,2.00>
      

    Total: 42560 W: 11177 L: 11023 D: 20360
      

    Ptnml(0-2): 118, 5092, 10727, 5204, 139
sprt @ 10+0.1 th 1
cores: 48 (6)
Reduce when probcut fails without a tt entry
26-01-18 jay
remove-prefetch-do_move diff 
  
  
    LLR: 0.58 (-2.94,2.94) <-1.75,0.25>
      

    Total: 20552 W: 5267 L: 5239 D: 10046
      

    Ptnml(0-2): 32, 2350, 5477, 2392, 25
sprt @ 5+0.05 th 8
cores: 107 (5)
Non regression SMP test with a larger hash size.
26-01-18 fau
toor3 diff 
  
  
    LLR: 0.53 (-2.94,2.94) <0.00,2.00>
      

    Total: 8640 W: 2215 L: 2156 D: 4269
      

    Ptnml(0-2): 18, 995, 2240, 1044, 23
sprt @ 10+0.1 th 1
cores: 16 (2)
toor3
26-01-18 Dan
32_b_keys diff 
  
  
    LLR: 0.25 (-2.94,2.94) <0.00,2.00>
      

    Total: 18176 W: 4782 L: 4732 D: 8662
      

    Ptnml(0-2): 62, 2180, 4551, 2236, 59
sprt @ 10+0.1 th 1
cores: 31 (5)
try 32 bit keys
26-01-18 fau
antid3 diff 
  
  
    LLR: 0.23 (-2.94,2.94) <0.00,2.00>
      

    Total: 10432 W: 2767 L: 2731 D: 4934
      

    Ptnml(0-2): 28, 1230, 2659, 1276, 23
sprt @ 10+0.1 th 1
cores: 13 (4)
antid3
26-01-14 Viz
prbShortcut1 diff 
  
  
    LLR: 0.23 (-2.94,2.94) <0.00,2.00>
      

    Total: 189728 W: 49071 L: 48772 D: 91885
      

    Ptnml(0-2): 533, 22368, 48824, 22545, 594
sprt @ 10+0.1 th 1
cores: 20 (3)
Take 1
26-01-18 fau
antid1 diff 
  
  
    LLR: 0.12 (-2.94,2.94) <0.00,2.00>
      

    Total: 10848 W: 2787 L: 2761 D: 5300
      

    Ptnml(0-2): 31, 1248, 2842, 1270, 33
sprt @ 10+0.1 th 1
cores: 19 (3)
antid1
26-01-17 sg
tuned_lmr_conditions2 diff 
  
  
    LLR: 0.10 (-2.94,2.94) <0.50,2.50>
      

    Total: 35490 W: 8993 L: 8913 D: 17584
      

    Ptnml(0-2): 11, 3848, 9954, 3914, 18
sprt @ 60+0.6 th 1
cores: 70 (6)
LTC: [Rebased] After 92k LTC lmr conditions tuning test most significant negative param A[0][1][2] (because less reduction scales good at ttPv nodes skip this condition).
26-01-18 Viz
longImpr2 diff 
  
  
    LLR: -0.04 (-2.94,2.94) <0.00,2.00>
      

    Total: 1600 W: 415 L: 416 D: 769
      

    Ptnml(0-2): 3, 201, 394, 198, 4
sprt @ 10+0.1 th 1
cores: 16 (2)
Take 2
26-01-16 sg
tuned_lmr_conditions2b diff 
  
  
    LLR: -0.10 (-2.94,2.94) <0.00,2.00>
      

    Total: 110208 W: 28604 L: 28450 D: 53154
      

    Ptnml(0-2): 332, 13013, 28304, 13079, 376
sprt @ 10+0.1 th 1
cores: 16 (2)
After 92k LTC lmr conditions tuning test most significant negative param A[0][1][2] but use opposite precondition ttPv (the original precondition failed and allowing both struggles but seem at least neutral).
26-01-17 sg
tuned_lmr_conditions2a diff 
  
  
    LLR: -0.17 (-2.94,2.94) <0.50,2.50>
      

    Total: 13146 W: 3321 L: 3308 D: 6517
      

    Ptnml(0-2): 2, 1415, 3733, 1414, 9
sprt @ 60+0.6 th 1
cores: 31 (6)
LTC: STC failed but as values were tuned at LTC try this now (TP 50%). After 92k LTC lmr conditions tuning test most significant negative param A[0][1][2].
26-01-18 Viz
longImpr1 diff 
  
  
    LLR: -0.26 (-2.94,2.94) <0.00,2.00>
      

    Total: 1664 W: 421 L: 441 D: 802
      

    Ptnml(0-2): 7, 197, 440, 185, 3
sprt @ 10+0.1 th 1
cores: 16 (2)
Take 1
26-01-16 sni
style diff 
  
  
    LLR: -0.27 (-2.94,2.94) <0.00,2.00>
      

    Total: 100736 W: 25901 L: 25777 D: 49058
      

    Ptnml(0-2): 285, 11847, 26020, 11891, 325
sprt @ 10+0.1 th 1
cores: 27 (4)
take 3
26-01-14 Viz
seerYoink2 diff 
  
  
    LLR: -0.28 (-2.94,2.94) <0.00,2.00>
      

    Total: 192576 W: 49921 L: 49664 D: 92991
      

    Ptnml(0-2): 538, 22516, 49947, 22725, 562
sprt @ 10+0.1 th 1
cores: 22 (3)
Take 2
26-01-15 sni
style diff 
  
  
    LLR: -0.34 (-2.94,2.94) <0.00,2.00>
      

    Total: 120288 W: 31170 L: 31023 D: 58095
      

    Ptnml(0-2): 368, 14233, 30806, 14358, 379
sprt @ 10+0.1 th 1
cores: 22 (4)
Tal style = 1 when simple*nnue > 0
26-01-16 Dan
smaller_si diff 
  
  
    LLR: -0.41 (-2.94,2.94) <0.00,2.00>
      

    Total: 116544 W: 30004 L: 29874 D: 56666
      

    Ptnml(0-2): 311, 12870, 31798, 12964, 329
sprt @ 10+0.1 th 1
cores: 24 (4)
speedup = +0.0021, P(speedup > 0) = 0.9744, i7-8750H
26-01-18 fau
toor3a diff 
  
  
    LLR: -0.82 (-2.94,2.94) <0.00,2.00>
      

    Total: 11296 W: 2856 L: 2912 D: 5528
      

    Ptnml(0-2): 33, 1348, 2934, 1308, 25
sprt @ 10+0.1 th 1
cores: 16 (2)
toor3a
26-01-18 fau
toor3b diff 
  
  
    LLR: -0.83 (-2.94,2.94) <0.00,2.00>
      

    Total: 10528 W: 2721 L: 2780 D: 5027
      

    Ptnml(0-2): 40, 1275, 2692, 1218, 39
sprt @ 10+0.1 th 1
cores: 18 (1)
toor3b
26-01-15 sg
tt_stm2 diff 
  
  
    LLR: -0.84 (-2.94,2.94) <0.00,2.00>
      

    Total: 131040 W: 33457 L: 33344 D: 64239
      

    Ptnml(0-2): 156, 14858, 35393, 14943, 170
sprt @ 5+0.05 th 8
cores: 48 (5)
STC gainer test with hash pressure struggles. So try STC SMP with hash pressure (4 MB) because theoretically the should be more collision because of the overlapping search of the different threads. Shift position key one bit up and replace lowest bit with side to move. So white and black side to move positions never matches as hit in TT.
26-01-15 sni
style^ diff 
  
  
    LLR: -0.86 (-2.94,2.94) <0.00,2.00>
      

    Total: 152960 W: 39495 L: 39346 D: 74119
      

    Ptnml(0-2): 477, 18101, 39192, 18216, 494
sprt @ 10+0.1 th 1
cores: 25 (4)
Tal style = 3 when simple*nnue > 0
26-01-17 sni
style diff 
  
  
    LLR: -0.92 (-2.94,2.94) <0.00,2.00>
      

    Total: 63616 W: 16432 L: 16420 D: 30764
      

    Ptnml(0-2): 199, 7528, 16335, 7554, 192
sprt @ 10+0.1 th 1
cores: 24 (3)
Tal style = 1 when simple*nnue < 0
26-01-12 sg
tt_stm2 diff 
  
  
    LLR: -1.11 (-2.94,2.94) <0.00,2.00>
      

    Total: 333504 W: 85646 L: 85253 D: 162605
      

    Ptnml(0-2): 1041, 39328, 85665, 39633, 1085
sprt @ 10+0.1 th 1
cores: 32 (4)
With normal hash pressure this change seems to have not much effect. So try with hash pressure (1 MB). Shift position key one bit up and replace lowest bit with side to move. So white and black side to move positions never matches as hit in TT.
26-01-12 Viz
pvLmrDeeS1 diff 
  
  
    LLR: -1.13 (-2.94,2.94) <0.50,2.50>
      

    Total: 53632 W: 13786 L: 13771 D: 26075
      

    Ptnml(0-2): 7, 4880, 17026, 4897, 6
sprt @ 60+0.6 th 8
cores: 195 (14)
LTC: ??
26-01-17 Viz
seerYoink3 diff 
  
  
    LLR: -1.81 (-2.94,2.94) <0.00,2.00>
      

    Total: 59008 W: 15152 L: 15225 D: 28631
      

    Ptnml(0-2): 153, 6985, 15315, 6884, 167
sprt @ 10+0.1 th 1
cores: 18 (3)
Take 3
26-01-15 sg
tt_stm2 diff 
  
  
    LLR: -1.84 (-2.94,2.94) <0.00,2.00>
      

    Total: 50682 W: 12877 L: 12959 D: 24846
      

    Ptnml(0-2): 36, 5527, 14278, 5483, 17
sprt @ 60+0.6 th 1
cores: 36 (6)
LTC: Regression tests with default hash passed and LTC with even around 1.76 elo. The hash pressure tests at low TC (singel and SMP) seems not good. So retest the non-regression lTC as LTC gainer with STC bounds and try the VVLTC route. Shift position key one bit up and replace lowest bit with side to move. So white and black side to move positions never match as hit in TT.
26-01-16 Dan
stat_ld_ph_1 diff 
  
  
    LLR: -2.36 (-2.94,2.94) <0.00,2.00>
      

    Total: 141664 W: 36458 L: 36459 D: 68747
      

    Ptnml(0-2): 439, 16765, 36410, 16794, 424
sprt @ 10+0.1 th 1
cores: 24 (3)
take 1
26-01-16 vdv
master diff 
  
  
    Elo: 46.32 ± 1.4 (95%) LOS: 100.0%
      

    Total: 53318 W: 17477 L: 10410 D: 25431
      

    Ptnml(0-2): 16, 2739, 14237, 9496, 171
    

    nElo: 100.88 ± 3.2 (95%) PairsRatio: 3.51
60000 @ 60+0.6 th 1
cores: 78 (5)
Progression test of "update the WDL model" of Jan 15 vs SF_17.
26-01-16 vdv
master diff 
  
  
    Elo: 29.61 ± 1.3 (95%) LOS: 100.0%
      

    Total: 52744 W: 16690 L: 12206 D: 23848
      

    Ptnml(0-2): 0, 3076, 15736, 7560, 0
    

    nElo: 68.27 ± 3.1 (95%) PairsRatio: 2.46
60000 @ 60+0.6 th 1
cores: 99 (9)
Endgames Progression test of "update the WDL model" of Jan 15 vs SF_17.
26-01-16 vdv
master diff 
  
  
    Elo: 58.27 ± 1.5 (95%) LOS: 100.0%
      

    Total: 49592 W: 15833 L: 7593 D: 26166
      

    Ptnml(0-2): 37, 2525, 11802, 10025, 407
    

    nElo: 119.43 ± 3.4 (95%) PairsRatio: 4.07
60000 @ 60+0.6 th 1
cores: 104 (7)
DFRC Progression test of "update the WDL model" of Jan 15 vs SF_17.
26-01-16 vdv
master diff 
  
  
    Elo: 43.69 ± 1.3 (95%) LOS: 100.0%
      

    Total: 47804 W: 15699 L: 9719 D: 22386
      

    Ptnml(0-2): 1, 1869, 14227, 7759, 46
    

    nElo: 104.18 ± 3.3 (95%) PairsRatio: 4.17
60000 @ 60+0.6 th 8
cores: 557 (8)
SMP Progression test of "update the WDL model" of Jan 15 vs SF_17.
26-01-16 fau
prova diff 
  
  
    20834/150000 iterations
      

    41668/300000 games played
300000 @ 60+0.6 th 1
cores: 34 (4)
prova - this should fix the "Infinite Extension" bug
26-01-17 Dan
cm_mc_b_1_tune diff 
  
  
    7547/30000 iterations
      

    15094/60000 games played
60000 @ 60+0.6 th 1
cores: 51 (5)
merge

Finished - 180146 tests

26-01-18 max feature/lazy-pawn-value diff 
  
  
    LLR: -2.95 (-2.94,2.94) <0.00,2.00>
      

    Total: 11120 W: 2725 L: 2960 D: 5435
      

    Ptnml(0-2): 22, 1362, 3020, 1141, 15
sprt @ 5+0.05 th 8 Add lazy value cache for pawn history reads Per-thread 4096-entry cache stores individual (pawnKey, pc, to) -> value mappings instead of copying entire 2KB entries. Each entry is 16 bytes (8 byte key + 3 byte pc/to + 2 byte value), total ~64KB fits in L1 cache. Uses lambda for lazy evaluation - atomic read from shared L1 only happens on cache miss. On cache hit, returns cached value (non-atomic, fast). Targets atomic READ contention in movepick (the actual bottleneck).
26-01-18 max feature/l0-pawn-cache-e diff 
  
  
    LLR: -2.94 (-2.94,2.94) <0.00,2.00>
      

    Total: 20512 W: 5104 L: 5324 D: 10084
      

    Ptnml(0-2): 36, 2419, 5546, 2239, 16
sprt @ 5+0.05 th 8 L0 pawn history cache for reduced atomic contention at high thread counts
26-01-18 kev probcut-depth diff 
  
  
    LLR: -2.94 (-2.94,2.94) <0.00,2.00>
      

    Total: 26400 W: 6782 L: 7005 D: 12613
      

    Ptnml(0-2): 98, 3225, 6751, 3054, 72
sprt @ 10+0.1 th 1 Reduce more in probcut based on our current depth
26-01-18 Dan loc_corr_2 diff 
  
  
    LLR: -1.10 (-2.94,2.94) <0.00,2.00>
      

    Total: 3736 W: 924 L: 1011 D: 1801
      

    Ptnml(0-2): 4, 465, 1016, 380, 3
sprt @ 5+0.05 th 8 try different linear combination
26-01-16 lem gcc-inline-100 diff 
  
  
    LLR: -1.95 (-2.94,2.94) <0.00,2.00>
      

    Total: 124800 W: 32165 L: 32155 D: 60480
      

    Ptnml(0-2): 350, 13805, 34130, 13715, 400
sprt @ 10+0.1 th 1 Positive results when raising inline limit for clang. See if same magic happens on GCC.
26-01-15 fau antid diff 
  
  
    29253/100000 iterations
      

    58506/200000 games played
200000 @ 60+0.6 th 1 antid
26-01-17 Dan loc_corr diff 
  
  
    LLR: -2.95 (-2.94,2.94) <0.00,2.00>
      

    Total: 14904 W: 3730 L: 3958 D: 7216
      

    Ptnml(0-2): 19, 1799, 4043, 1573, 18
sprt @ 5+0.05 th 8 try average local corrhists with shared corrhists
26-01-16 lem gcc-inline-200 diff 
  
  
    LLR: -2.93 (-2.94,2.94) <0.00,2.00>
      

    Total: 114816 W: 29335 L: 29423 D: 56058
      

    Ptnml(0-2): 337, 12678, 31462, 12598, 333
sprt @ 10+0.1 th 1 Positive results when raising inline limit for clang. See if same magic happens on GCC.
26-01-14 Fis GCCProfile diff 
  
  
    LLR: -3.24 (-2.94,2.94) <0.00,2.00>
      

    Total: 297952 W: 76789 L: 76643 D: 144520
      

    Ptnml(0-2): 883, 32949, 81178, 33071, 895
sprt @ 10+0.1 th 1 Extra profile directives for GCC Inspired by https://tests.stockfishchess.org/tests/view/696667bdd5a3b5895b50f3f2
26-01-18 max feature/twolevel-pawn-h diff 
  
  
    LLR: -0.26 (-2.94,2.94) <0.00,2.00>
      

    Total: 384 W: 87 L: 110 D: 187
      

    Ptnml(0-2): 0, 60, 97, 33, 2
sprt @ 10+0.1 th 1 Adaptive L0 pawn history cache size based on thread count L0 cache scales with thread count: 512 * numThreads, capped at 8192 (full size). This reduces evictions and L1 atomic accesses at higher thread counts. At 16+ threads each worker gets a full-size L0 cache, minimizing shared L1 contention. Single-thread mode bypasses L0/L1 entirely, using a dedicated non-atomic table.
26-01-16 ane split-ttmovehist diff 
  
  
    LLR: -0.74 (-2.94,2.94) <0.00,2.00>
      

    Total: 141344 W: 36245 L: 36103 D: 68996
      

    Ptnml(0-2): 463, 16641, 36266, 16895, 407
sprt @ 10+0.1 th 1 Take 2, try adjusting bonus values to compensate for splitting
26-01-17 max feature/embedded-tableb diff 
  
  
    LLR: -2.99 (-2.94,2.94) <0.00,2.00>
      

    Total: 6784 W: 1665 L: 1920 D: 3199
      

    Ptnml(0-2): 43, 861, 1816, 652, 20
sprt @ 10+0.1 th 1 Embedded O(1) compressed 3-4 piece tablebases
26-01-12 Viz lesserYoink7 diff 
  
  
    LLR: -2.94 (-2.94,2.94) <0.00,2.00>
      

    Total: 293056 W: 75358 L: 75193 D: 142505
      

    Ptnml(0-2): 792, 32744, 79323, 32845, 824
sprt @ 10+0.1 th 1 Take 7
26-01-17 jay remove-givescheck-param diff 
  
  
    LLR: -0.43 (-2.94,2.94) <-1.75,0.25>
      

    Total: 44448 W: 11379 L: 11463 D: 21606
      

    Ptnml(0-2): 117, 4973, 12138, 4869, 127
sprt @ 10+0.1 th 1 remove the givesCheck param in do_move get rid of a few vars
26-01-16 jay qsearch-ttmove diff 
  
  
    LLR: -1.83 (-2.94,2.94) <0.00,2.00>
      

    Total: 110624 W: 28608 L: 28608 D: 53408
      

    Ptnml(0-2): 368, 12988, 28591, 13006, 359
sprt @ 10+0.1 th 1 qsearch: exclude tt from pruning
26-01-17 fau toor2a diff 
  
  
    LLR: -1.40 (-2.94,2.94) <0.00,2.00>
      

    Total: 22592 W: 5814 L: 5906 D: 10872
      

    Ptnml(0-2): 70, 2757, 5739, 2655, 75
sprt @ 10+0.1 th 1 toor2a
26-01-17 max feature/pawn-history-la diff 
  
  
    LLR: -0.01 (-2.94,2.94) <-1.75,0.25>
      

    Total: 1408 W: 371 L: 373 D: 664
      

    Ptnml(0-2): 3, 162, 377, 158, 4
sprt @ 10+0.1 th 1 Thread-local pawn history with large page allocation (2MB per thread, 1024 buckets)
26-01-17 fau toor2 diff 
  
  
    LLR: -2.45 (-2.94,2.94) <0.00,2.00>
      

    Total: 29632 W: 7541 L: 7715 D: 14376
      

    Ptnml(0-2): 98, 3587, 7609, 3435, 87
sprt @ 10+0.1 th 1 toor2
26-01-17 fau toor1 diff 
  
  
    LLR: -2.95 (-2.94,2.94) <0.00,2.00>
      

    Total: 22304 W: 5590 L: 5818 D: 10896
      

    Ptnml(0-2): 75, 2734, 5731, 2568, 44
sprt @ 10+0.1 th 1 toor1
26-01-11 max feature/qsearch-repetit diff 
  
  
    LLR: -0.82 (-2.94,2.94) <-1.75,0.25>
      

    Total: 190758 W: 48629 L: 48890 D: 93239
      

    Ptnml(0-2): 109, 20881, 53637, 20666, 86
sprt @ 60+0.6 th 1 LTC: Qsearch repetition guard simplification
26-01-17 fau toor diff 
  
  
    LLR: -2.93 (-2.94,2.94) <0.00,2.00>
      

    Total: 20672 W: 5232 L: 5464 D: 9976
      

    Ptnml(0-2): 72, 2577, 5264, 2357, 66
sprt @ 10+0.1 th 1 toor
26-01-16 max feature/lowply-simplify diff 
  
  
    LLR: -0.04 (-2.94,2.94) <-1.75,0.25>
      

    Total: 67680 W: 17358 L: 17436 D: 32886
      

    Ptnml(0-2): 210, 8067, 17338, 8041, 184
sprt @ 10+0.1 th 1 Low ply history simplification
26-01-14 Viz prbShortcut3 diff 
  
  
    LLR: -2.94 (-2.94,2.94) <0.00,2.00>
      

    Total: 91744 W: 23505 L: 23631 D: 44608
      

    Ptnml(0-2): 263, 10951, 23553, 10859, 246
sprt @ 10+0.1 th 1 Take 3
26-01-15 lem raise-inline-threshold diff 
  
  
    LLR: 2.97 (-2.94,2.94) <-1.75,0.25>
      

    Total: 128224 W: 33008 L: 32894 D: 62322
      

    Ptnml(0-2): 258, 13572, 36305, 13752, 225
sprt @ 10+0.1 th 1 Try https://tests.stockfishchess.org/tests/view/696884a56d118e46e1731473 on apple silicon
26-01-17 sni style diff 
  
  
    LLR: -2.42 (-2.94,2.94) <0.00,2.00>
      

    Total: 15168 W: 3855 L: 4051 D: 7262
      

    Ptnml(0-2): 73, 1886, 3835, 1744, 46
sprt @ 10+0.1 th 1 Tal style = 5