Stockfish Testing Queue

Pending - 8 tests 22.9 hrs

11-06-16 aj master diff
ELO: 33.48 +-2.7 (95%) LOS: 100.0%
Total: 13356 W: 2129 L: 846 D: 10381
40000 @ 20+0.2 th 7 SMP Regression test 1. First SMP Regression test since SF-7 release 2. Framework is mostly empty
23-06-16 sn drawish diff
LLR: -1.93 (-2.94,2.94) [0.00,5.00]
Total: 45611 W: 8639 L: 8521 D: 28451
sprt @ 10+0.1 th 1 Positions with equal material and compact symmetrical pawn chains are drawish
21-06-16 mb threads_at_root1 diff
LLR: -0.60 (-2.94,2.94) [0.00,5.00]
Total: 17000 W: 2864 L: 2821 D: 11315
sprt @ 5+0.05 th 7 Ignore fail highs at root for deeper threads.
29-06-16 sn tune_komodo diff
6991/15000 iterations
14119/30000 games played
30000 @ 10+0.1 th 1 Tune with fixed optimism_mobility=10
28-06-16 pb circular_clear_cm diff
LLR: -2.54 (-2.94,2.94) [0.00,5.00]
Total: 11000 W: 1787 L: 1849 D: 7364
sprt @ 5+0.05 th 7 STC: Clear countermoves of helper-threads with a round robin
26-06-16 SC LMRstatsTuning diff
25016/50000 iterations
50486/100000 games played
100000 @ 10+0.1 th 1 I dont have the slightest idea of whether the parameters I chose are ok, but it was not that bad. So try to tune it. Restart with higher c. Will stop if it converges early.
24-06-16 sg quience_search diff
LLR: -2.74 (-2.94,2.94) [-3.00,1.00]
Total: 102694 W: 19257 L: 19585 D: 63852
sprt @ 10+0.1 th 1 remove recapture phase from quience search move generation
24-06-16 pb halfdensity_8block diff
ELO: -5.69 +-4.3 (95%) LOS: 0.5%
Total: 8000 W: 1232 L: 1363 D: 5405
20000 @ 3+0.03 th 21 Agreed with Peter Zsifkovits (CoffeeOne) to do some tests on large number of threads. First one is to verify if the last block in the halfdensitymap (skipsize=4) brings any benefit.

Active - 57 machines 224 cores 1.48M nps (332.13M total nps) 372 games/minute

Machine Cores MNps System Version Running on Last updated
mgrabiak
3
1.53 Windows 8 59 ZeroDepthPruning seconds ago
cw
3
0.95 Windows 10 59 ZeroDepthPruning seconds ago
teddybaer
7
2.03 Linux 3.16.0-4-amd64 59 ZeroDepthPruning seconds ago
mibere
14
1.58 Linux 3.16.0-4-amd64-2 59 ZeroDepthPruning seconds ago
nssy
3
2.14 Linux 3.12.14-srv-nssy4 59 ZeroDepthPruning seconds ago
Bobo1239
1
0.00 Linux 4.6.2-1-ARCH 59 ZeroDepthPruning seconds ago
ttruscott
3
1.69 Windows 7 59 ZeroDepthPruning seconds ago
mibere
7
1.19 Linux 3.16.0-4-amd64-7 59 ZeroDepthPruning seconds ago
mibere
5
1.74 Linux 3.16.0-4-amd64-1 59 ZeroDepthPruning seconds ago
Freja
3
1.72 Darwin 15.4.0 59 end_draw seconds ago
cw
5
0.75 Windows 7 59 end_draw seconds ago
Fisherman
6
1.62 Windows 7 59 end_draw seconds ago
mibere
7
1.54 Linux 3.16.0-4-amd64-8 59 end_draw seconds ago
CSU_Dynasty
11
1.27 Windows 7 59 end_draw seconds ago
ctoks
3
2.39 Windows 8.1 59 end_draw seconds ago
amicic
3
1.99 Windows 7 59 end_draw seconds ago
spams
3
2.42 Windows 7 59 end_draw seconds ago
cw
5
0.74 Windows 7 59 splitMobilityTune 40 minutes ago
ctoks
2
1.16 Windows 8.1 59 splitMobilityTune seconds ago
mhoram
2
0.76 Linux 3.16.0-4-amd64 59 komodo seconds ago
davar
3
1.78 Windows 8 59 komodo seconds ago
leszek
7
0.97 Windows 2003Server 59 komodo seconds ago
mibere
3
1.11 Linux 3.16.0-4-amd64-4 59 komodo seconds ago
marrco
7
1.21 Windows 2012Server 59 tuneTMM seconds ago
CSU_Dynasty
7
0.84 Windows 7 59 tuneTMM seconds ago
modolief
1
1.68 Windows 8 59 tuneTMM seconds ago
cw
3
1.43 Windows 7 59 tuneTMM seconds ago
mibere
7
0.93 Linux 3.16.0-4-amd64-6 59 tuneTMM seconds ago
horst.prack
1
1.29 Linux 4.6.0-1-amd64 59 tuneTMM seconds ago
crunchy
7
1.41 Windows 7 59 MoreCheck seconds ago
bioMatrix
3
0.99 Windows 8 59 MoreCheck 11 minutes ago
ako027ako
3
1.90 Windows 10 59 MoreCheck seconds ago
Nesa92
3
1.43 Windows 7 59 MoreCheck seconds ago
mibere
7
1.45 Linux 3.16.0-4-amd64-3 59 MoreCheck seconds ago
mibere
7
1.40 Linux 3.16.0-4-amd64-5 59 MoreCheck seconds ago
teddybaer
5
2.14 Linux 3.16.0-4-amd64 59 MoreCheck seconds ago
sunu
2
1.44 Linux 4.6.0-1.slh.1-aptosid-amd64 59 MoreCheck seconds ago
drabel
3
2.19 Windows post2008Server 59 MoreCheck seconds ago
master_oogway
7
1.09 Linux 4.5.0-0.bpo.2-amd64 59 MoreCheck 52 minutes ago
psk
3
2.49 Linux 3.13.0-88-generic 59 MoreCheck seconds ago
praetoriansentry
6
2.24 Windows 10 59 MoreCheck seconds ago
stocky
3
2.16 Linux 3.13.0-87-generic 59 futilityPawns seconds ago
velislav
2
1.60 Linux 3.10.0-123.8.1.el7.x86_64 59 futilityPawns seconds ago
vdbergh
5
1.46 Linux 2.6.32-504.23.4.el6.x86_64 59 futilityPawns seconds ago
spams
3
1.97 Windows 7 59 futilityPawns seconds ago
praetoriansentry
2
1.69 Windows 10 59 futilityPawns seconds ago
SC
3
2.13 Linux 4.2.0-38-generic 59 futilityPawns seconds ago
velislav
1
1.47 Linux 3.13.0-86-generic 59 futilityPawns seconds ago
homyur
2
2.05 Windows 7 59 futilityPawns seconds ago
cw
3
1.20 Windows 7 59 futilityPawns seconds ago
fp53fish
1
1.10 Linux 3.19.0-32-generic 59 futilityPawns 2 minutes ago
biffhero
1
1.28 Linux 3.16.0-4-amd64 59 futilityPawns 3 minutes ago
cw
1
1.80 Windows 7 59 futilityPawns seconds ago
sergeballif
1
1.24 Windows 7 59 futilityPawns 2 minutes ago
drabel
1
1.34 Windows 8 59 futilityPawns 2 minutes ago
drabel
1
1.37 Windows 8 59 futilityPawns 3 minutes ago
cw
3
1.17 Windows 7 59 mult_stop seconds ago
29-06-16 El end_draw diff
LLR: -0.26 (-2.94,2.94) [0.00,5.00]
Total: 3884 W: 559 L: 556 D: 2769
sprt @ 10+0.1 th 1 My take on drawish endgames with equal material and compact pawn chains.
29-06-16 SC ZeroDepthPruning diff
LLR: -1.56 (-2.94,2.94) [-3.00,1.00]
Total: 4584 W: 837 L: 929 D: 2818
sprt @ 10+0.1 th 1 A variation of the previous ParentNodePruning idea. Will see whether this or the previous one is better, tune it and then have a last go.
29-06-16 Ro MoreCheck diff
LLR: -0.72 (-2.94,2.94) [0.00,5.00]
Total: 10487 W: 1446 L: 1439 D: 7602
sprt @ 60+0.6 th 1 LTC for this king safety patch which was yellow at STC
28-06-16 jo tuneTMM diff
11692/20000 iterations
23706/40000 games played
40000 @ 40+0.4 th 1 Retune time management at 4 x longer tc. SF sometimes moves too fast, imho. Do we get different values? (Half throughput.)
29-06-16 sn komodo diff
ELO: -2.11 +-3.8 (95%) LOS: 14.0%
Total: 13192 W: 2694 L: 2774 D: 7724
20000 @ 10+0.1 th 1 Take 5bis: half values of take 5
27-06-16 SC futilityPawns diff
LLR: -0.97 (-2.94,2.94) [-3.00,1.00]
Total: 71216 W: 9695 L: 9841 D: 51680
sprt @ 60+0.6 th 1 Prune futile nodes also if we have only pawns. Was +2 ELO after 1000 local games, so for SPRT. LTC. Fixed hash.
29-06-16 Fi splitMobilityTune diff
1935/20000 iterations
3922/40000 games played
40000 @ 20+0.2 th 1 Split mobility into lower and upper ranks. Tune.
26-06-16 pe mult_stop diff
LLR: -2.25 (-2.94,2.94) [0.00,5.00]
Total: 38212 W: 4904 L: 4871 D: 28437
sprt @ 40+0.4 th 3 LTC Tuned values

Finished - 11818 tests

28-06-16 sn komodo diff
ELO: -2.55 +-3.2 (95%) LOS: 5.9%
Total: 20000 W: 4334 L: 4481 D: 11185
20000 @ 10+0.1 th 1 Take 5: try the resulting values of tuning #3 (with fixed OPTIMISM_PAWNS=10 for us)
29-06-16 ci prior_killer diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17378 W: 3225 L: 3275 D: 10878
sprt @ 10+0.1 th 1 Updating killer moves after fail low.
28-06-16 sn tropism2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 19752 W: 3736 L: 3775 D: 12241
sprt @ 10+0.1 th 1 King tropism using double attacks, take 2.
28-06-16 Vo clearRCM diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17334 W: 3248 L: 3298 D: 10788
sprt @ 10+0.1 th 1 Clear refuted cm.
28-06-16 SC ParentNodePruning diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 15553 W: 2834 L: 3019 D: 9700
sprt @ 10+0.1 th 1 Is it possible to unify razoring and futility pruning? An exotic, locally tuned attempt.
28-06-16 sn komodo diff
ELO: 2.26 +-2.9 (95%) LOS: 93.4%
Total: 20000 W: 3801 L: 3671 D: 12528
20000 @ 10+0.1 th 1 Take 4bis: half values of take 4
28-06-16 sn komodo diff
ELO: -0.47 +-2.9 (95%) LOS: 37.6%
Total: 20000 W: 3641 L: 3668 D: 12691
20000 @ 10+0.1 th 1 Take 4: try the resulting value of tuning #2 (with fixed OPTIMISM_PIECES=10 for us)
27-06-16 sn tune_komodo diff
14772/15000 iterations
30000/30000 games played
30000 @ 10+0.1 th 1 Tune with fixed optimism_pawns=10
28-06-16 pb circular_clear_cm diff
ELO: 4.04 +-4.6 (95%) LOS: 95.9%
Total: 8000 W: 1485 L: 1392 D: 5123
8000 @ 3+0.03 th 7 Clear countermoves of helper-threads with a round robin
28-06-16 My ms diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8893 W: 1622 L: 1709 D: 5562
sprt @ 10+0.1 th 1 cap opp bishops material
25-06-16 pb lazy_big_map diff
ELO: 0.59 +-3.7 (95%) LOS: 62.2%
Total: 10000 W: 1504 L: 1487 D: 7009
10000 @ 3+0.03 th 44 Can we further extend the halfdensity map? Quick check.
28-06-16 Vo cmDepth' diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10255 W: 1865 L: 1947 D: 6443
sprt @ 10+0.1 th 1 Don't overwrite cms at lower depths.
28-06-16 SC ParentNodePruning diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 2956 W: 489 L: 653 D: 1814
sprt @ 10+0.1 th 1 Is it possible to unify razoring and futility pruning? An exotic attempt, but one never knows. Low throughput and fixed bench.
27-06-16 sn tune_komodo diff
14766/15000 iterations
30000/30000 games played
30000 @ 10+0.1 th 1 Tune with fixed optimism_pieces=10
27-06-16 sn komodo diff
ELO: -2.54 +-3.0 (95%) LOS: 4.9%
Total: 20000 W: 3843 L: 3989 D: 12168
20000 @ 10+0.1 th 1 Take 3: explore the effect of negative material optimism for them. Rescheduled with fixed number of games (20000) to compare progress since take 2.
27-06-16 Vo statAgreement diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8737 W: 1616 L: 1704 D: 5417
sprt @ 10+0.1 th 1 Version B.
27-06-16 sn tropism2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5295 W: 1010 L: 1114 D: 3171
sprt @ 10+0.1 th 1 King tropism, using double attacks
26-06-16 Ro MoreCheck diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 55453 W: 10431 L: 10315 D: 34707
sprt @ 10+0.1 th 1 Take 3: no contact checks and no case 2
26-06-16 mc nmp_staticeval2 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 52346 W: 7167 L: 7109 D: 38070
sprt @ 60+0.6 th 1 LTC: Don't NMP if staticEval < beta: LTC
27-06-16 Vo statAgreement diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6948 W: 1281 L: 1377 D: 4290
sprt @ 10+0.1 th 1 Don't use history score if value disagrees with cm and fms values.
27-06-16 sn komodo diff
LLR: 0.17 (-2.94,2.94) [0.00,5.00]
Total: 142 W: 30 L: 22 D: 90
sprt @ 10+0.1 th 1 Take 3: explore the effect of negative material optimism for them
27-06-16 sn tune_komodo diff
14841/15000 iterations
30000/30000 games played
30000 @ 10+0.1 th 1 Re-tune all three parameters (without the bug)
27-06-16 sn komodo diff
ELO: -1.16 +-2.9 (95%) LOS: 21.8%
Total: 20000 W: 3673 L: 3740 D: 12587
20000 @ 10+0.1 th 1 Check if there is progress compared to take 1 with the result of tuning
26-06-16 SC futilityWin diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 41084 W: 7811 L: 7756 D: 25517
sprt @ 10+0.1 th 1 Try to use same logic as razoring in FP in order to resolve also cases where eval > VALUE_KNOWN_WIN
27-06-16 ci goodQuiet diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16130 W: 3023 L: 3078 D: 10029
sprt @ 10+0.1 th 1 Always sort moves, when all quiet moves are bad.
27-06-16 Fi blockedPawnInitiative diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 9537 W: 1718 L: 1803 D: 6016
sprt @ 10+0.1 th 1 Take 3. Penalize if none or just one pawn is unblocked.
26-06-16 II reduction_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 21011 W: 3959 L: 3993 D: 13059
sprt @ 10+0.1 th 1 One logical reduction tweak.
27-06-16 SC futilityPawns diff
LLR: 0.02 (-2.94,2.94) [-3.00,1.00]
Total: 29 W: 5 L: 4 D: 20
sprt @ 60+0.6 th 1 Prune futile nodes also if we have only pawns. Was +2 ELO after 1000 local games, so for SPRT. LTC.
26-06-16 SC futilityPawns diff
LLR: 3.29 (-2.94,2.94) [-3.00,1.00]
Total: 28407 W: 5336 L: 5210 D: 17861
sprt @ 10+0.1 th 1 Prune futile nodes also if we have only pawns. Was +2 ELO after 1000 local games, so for SPRT.
26-06-16 Vo pvfb diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 25446 W: 4722 L: 4737 D: 15987
sprt @ 10+0.1 th 1 Last try at this...
26-06-16 jo tmm_tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 3014 W: 530 L: 644 D: 1840
sprt @ 10+0.1 th 1 Take 2.
25-06-16 sn komodo diff
ELO: -14.03 +-3.4 (95%) LOS: 0.0%
Total: 20000 W: 4525 L: 5332 D: 10143
20000 @ 10+0.1 th 1 First draft at Komodo-style contempt: value our pieces, our pawns and our mobility a little bit more than the opponent's. Estimate the size of the Elo loss against current master.
25-06-16 aj main_history diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18465 W: 2738 L: 2791 D: 12936
sprt @ 10+0.1 th 7 After every iteration align with main thread history. Assumption is that main thread has the best history due to minimal skipping: STC
26-06-16 Fi blockedPawnInitiative diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 11707 W: 2186 L: 2261 D: 7260
sprt @ 10+0.1 th 1 Take 2
25-06-16 jo smp_mcp1 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13318 W: 2164 L: 2234 D: 8920
sprt @ 5+0.05 th 7 Less movecount based pruning in smp mode.
25-06-16 pe mult_stop diff
LLR: 2.94 (-2.94,2.94) [0.00,5.00]
Total: 36614 W: 6189 L: 5921 D: 24504
sprt @ 10+0.1 th 3 Tuned values
26-06-16 Fi blockedPawnInitiative diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 35356 W: 6614 L: 6586 D: 22156
sprt @ 10+0.1 th 1 Use blocked pawns in initiative.
25-06-16 SC LMRstatsTuning diff
8472/50000 iterations
17344/100000 games played
100000 @ 10+0.1 th 1 I dont have the slightest idea of whether the parameters I chose are ok, but it was not that bad. So try to tune it.
25-06-16 ci update_64 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 24620 W: 4577 L: 4649 D: 15394
sprt @ 10+0.1 th 1 Tweak to stats update factor.
25-06-16 sn tune_komodo diff
19618/20000 iterations
40000/40000 games played
40000 @ 10+0.1 th 1 Try to tune the Komodo-style contempt values...
26-06-16 Ro MoreCheck diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 9804 W: 1836 L: 1919 D: 6049
sprt @ 10+0.1 th 1 More safe checks for rooks and minors, from supported squares which are defended only once (by a queen or a king)
26-06-16 Vo pvBonus diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13741 W: 2595 L: 2661 D: 8485
sprt @ 10+0.1 th 1 Increase stats bonus for PvNodes
25-06-16 Vo cefp2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 16436 W: 3048 L: 3102 D: 10286
sprt @ 10+0.1 th 1 Take 2
25-06-16 Vo cefp diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 5899 W: 1072 L: 1173 D: 3654
sprt @ 10+0.1 th 1 Don't extend checks if value is way below alpha.
25-06-16 SC LMRstats diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13958 W: 2565 L: 2630 D: 8763
sprt @ 10+0.1 th 1 If best value is exceeding the mean value of searched moves of something around 3 sigma, and move count is larger than 5: increase reductions. Take 2, use LMR count and update with full depth value if available.
25-06-16 Ro TestSafe diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 14752 W: 2751 L: 2813 D: 9188
sprt @ 10+0.1 th 1 allow a pawn to support a pawn against a pawn when evaluating pawn push threat or pawn safe threats.
25-06-16 Vo lmrt diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 8442 W: 1499 L: 1589 D: 5354
sprt @ 10+0.1 th 1 lmr tweak
25-06-16 sn knight_vs_bishop diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10224 W: 1377 L: 1461 D: 7386
sprt @ 60+0.6 th 1 LTC: Knights are best with compact pawn structures
25-06-16 ci pc_bonus diff
LLR: 3.10 (-2.94,2.94) [-3.00,1.00]
Total: 25004 W: 3512 L: 3390 D: 18102
sprt @ 60+0.6 th 1 LTC: Small simplification for prior countermove bonus.
25-06-16 sn knight_vs_bishop diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 5525 W: 1103 L: 953 D: 3469
sprt @ 10+0.1 th 1 Knights are best with compact pawn structures