Stockfish Testing Queue

Finished - 22407 tests

04-12-14 sg aspiration_window diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 21378 W: 3560 L: 3560 D: 14258
sprt @ 60+0.05 th 1 LTC: try opposite. if NOT fail highs and lows alternate use tighter window (Take 4)
04-12-14 pe dbe329f diff
LLR: 2.95 (-2.94,2.94) [-1.00,4.00]
Total: 24845 W: 5148 L: 4950 D: 14747
sprt @ 15+0.05 th 1 Larger dc. High hash pressure
03-12-14 sg aspiration_window diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 47917 W: 9749 L: 9509 D: 28659
sprt @ 15+0.05 th 1 try opposite. if NOT fail highs and lows alternate use tighter window (Take 4)
03-12-14 sg aspiration_window diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 17535 W: 3480 L: 3539 D: 10516
sprt @ 15+0.05 th 1 counter search instability: if fail highs and lows alternate widen the window by don't changing opposite bound
03-12-14 sg aspiration_window diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 33826 W: 6851 L: 6864 D: 20111
sprt @ 15+0.05 th 1 try opposite. if NOT fail highs and lows alternate use tighter window (Take 3)
03-12-14 sg aspiration_window diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 11685 W: 2301 L: 2376 D: 7008
sprt @ 15+0.05 th 1 counter search instability: if fail highs and lows alternate widen the window (change opposite bound less) (Take 2)
03-12-14 sn rook3 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 7417 W: 1439 L: 1526 D: 4452
sprt @ 15+0.05 th 1 Decrease RookValueMg and RookValueEg by 15
02-12-14 gl piece_value_spsa diff
47639/50000 iterations
100000/100000 games played
100000 @ 15+0.05 th 1 SPSA tuning for piece values (modified ck, rk values based on initial parameter value)
02-12-14 Fi TT_sprats2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 7914 W: 1531 L: 1616 D: 4767
sprt @ 15+0.05 th 1 Increase number of TT entries by 33% by only storing eval value if either value or move are not storing anything which is 75% of the time. Inspired by TT_sprats by sf-x http://tests.stockfishchess.org/tests/view/543ab7420ebc59106cb96d81 I don't know enough about the search to know whether any changes there(like what sf-x did) would help this patch further. Help in that area is welcome.
02-12-14 jo piece_value_spsa diff
19673/20000 iterations
39970/40000 games played
40000 @ 15+0.05 th 1 Verification test of SPSA, set queen values to around a rook, and see what we get. Reschedule with larger ck and rk. 20k iterations only.
01-12-14 gl piece_value_spsa diff
39185/40000 iterations
80000/80000 games played
80000 @ 15+0.05 th 1 Verification test of SPSA, set queen values to around a rook, and see what we get
02-12-14 jo spsa_piece_values diff
ELO: -3.91 +-3.1 (95%) LOS: 0.6%
Total: 20000 W: 3964 L: 4189 D: 11847
20000 @ 15+0.05 th 1 Just a quick measurement of new material values.
01-12-14 pe dyn_cont diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 67656 W: 11550 L: 11399 D: 44707
sprt @ 60+0.05 th 1 LTC. As original objection was that dc works only in specific version of SF, test dc in this newer and much different version. Hash interaction may be important here, so I test with high pressure TCEC style hash.
30-11-14 gl master diff
ELO: 36.62 +-1.8 (95%) LOS: 100.0%
Total: 40000 W: 7962 L: 3761 D: 28277
40000 @ 60+0.05 th 3 3 threads regression test against sf5 using 8moves_v3
01-12-14 jo Outposts2 diff
LLR: -2.95 (-2.94,2.94) [-0.50,4.00]
Total: 20895 W: 4104 L: 4179 D: 12612
sprt @ 15+0.05 th 1 Formula based values tuned with SPSA.
01-12-14 Ro RelativePins diff
ELO: -8.37 +-3.0 (95%) LOS: 0.0%
Total: 20000 W: 3732 L: 4214 D: 12054
20000 @ 15+0.05 th 1 Measuring if any value to reducing mobility due to a pin against a Queen
01-12-14 tt copy_tt diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 24922 W: 5011 L: 5049 D: 14862
sprt @ 15+0.05 th 1 make local copy of tt entry in search stack. previously, the members of the tt entry that are read in the moves_loop could have been overwritten while the search is running.
01-12-14 sg king_safety diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 8478 W: 1693 L: 1777 D: 5008
sprt @ 15+0.05 th 1 use quadratic kingAttackersWeight
01-12-14 gl piece_value_tuned diff
LLR: -2.95 (-2.94,2.94) [-1.00,3.00]
Total: 5913 W: 1145 L: 1295 D: 3473
sprt @ 15+0.05 th 1 SPSA tuned piece values
29-11-14 gl master diff
ELO: 39.02 +-1.9 (95%) LOS: 100.0%
Total: 40000 W: 8624 L: 4151 D: 27225
40000 @ 60+0.05 th 1 Regression test against sf5 using 8moves_v3, previous was 36 +- 1.9
29-11-14 gl piece_value_spsa diff
95683/100000 iterations
199698/200000 games played
200000 @ 15+0.05 th 1 SPSA tuning for piece values
01-12-14 pe dyn_cont diff
LLR: 2.96 (-2.94,2.94) [-1.00,4.00]
Total: 5896 W: 1262 L: 1117 D: 3517
sprt @ 15+0.05 th 1 As original objection was that dc works only in specific version of SF, test dc in this newer and much different version. Hash interaction may be important here, so I test with high pressure TCEC style hash.
01-12-14 Ro KingRingShape diff
LLR: -2.99 (-2.94,2.94) [-1.50,4.50]
Total: 43045 W: 8667 L: 8656 D: 25722
sprt @ 15+0.05 th 1 See if any value to adding just one square behind the currently used KingRing. (e.g. If BLACK Kg6, we consider also the square g8)
30-11-14 Ro KingRingShape diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 27688 W: 5584 L: 5614 D: 16490
sprt @ 15+0.05 th 1 New shape for the KingRing used for King Safety evaluation. Precalculated in bitboard::init().
01-12-14 Ro KingRingShape diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 4655 W: 894 L: 988 D: 2773
sprt @ 15+0.05 th 1 OK, potential ELO gain here...now running SPRT 15
30-11-14 Ro KingRingShape diff
ELO: 2.40 +-3.0 (95%) LOS: 94.1%
Total: 20000 W: 3983 L: 3845 D: 12172
20000 @ 15+0.05 th 1 Take 3: See if any value in adding 2 step horizontals to current KingRing used in King Safety calculations
30-11-14 Ro KingRingShape diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 25661 W: 5095 L: 5131 D: 15435
sprt @ 15+0.05 th 1 Retry on take 1, same shape with knight distance squares added, but excluding squares behind the King. Measure value.
30-11-14 Ro KingRingShape diff
ELO: -3.77 +-3.1 (95%) LOS: 0.8%
Total: 20000 W: 3993 L: 4210 D: 11797
20000 @ 15+0.05 th 1 Take 2: add 2 step diagonals.
29-11-14 Ro MobilityProb diff
ELO: 0.35 +-3.1 (95%) LOS: 58.8%
Total: 20000 W: 4056 L: 4036 D: 11908
20000 @ 15+0.05 th 1 Take 2: Consider only x-ray through opponent Queens
29-11-14 jo clop-matimb3 diff
ELO: -10.40 +-3.0 (95%) LOS: 0.0%
Total: 24594 W: 5500 L: 6236 D: 12858
40000 @ 7+0.05 th 1 Measure set #2.
29-11-14 jo clop-matimb3 diff
ELO: -18.71 +-5.5 (95%) LOS: 0.0%
Total: 7137 W: 1502 L: 1886 D: 3749
40000 @ 7+0.05 th 1 Measure some new values for pawn, knight and bishop imbalance.
29-11-14 sg lmr diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 19180 W: 3834 L: 3888 D: 11458
sprt @ 15+0.05 th 1 less reduction if move has same target square as ttMove
29-11-14 Ro MobilityProb diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 16097 W: 3177 L: 3240 D: 9680
sprt @ 15+0.05 th 1 Trying to find the future mobility of some pieces (in cases of attack from a slider on higher value piece). For example, if our Bishop attacks a Rook, most likely the Rook will move away from this attack, which will result in a better mobility for our Bishop.
28-11-14 rn simpleprng diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 55481 W: 9202 L: 9446 D: 36833
sprt @ 60+0.05 th 1 LTC: Use a simpler PRNG and simplify how it is used. It also helps finding magics quicker. Testing as simplification.
27-11-14 mb reduced_attackunits diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 6855 W: 1086 L: 1172 D: 4597
sprt @ 60+0.05 th 1 Reduce attackUnits if only one piece is attacking. LTC because kingsafety is time dependent.
28-11-14 rn simpleprng diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 35515 W: 7077 L: 6982 D: 21456
sprt @ 15+0.05 th 1 Use a simpler PRNG and simplify how it is used. It also helps finding magics quicker. Testing as simplification.
27-11-14 Ro LowSafeMove2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 1854 W: 330 L: 432 D: 1092
sprt @ 15+0.05 th 1 Another try with smaller penalty
27-11-14 Ro MobilityArea diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 7883 W: 1550 L: 1635 D: 4698
sprt @ 15+0.05 th 1 Take 2: do not extend the Queen mobility.
27-11-14 jo Outposts2 diff
ELO: -6.07 +-2.7 (95%) LOS: 0.0%
Total: 30000 W: 6642 L: 7166 D: 16192
30000 @ 7+0.05 th 1 Since queue is empty, allow me one last try, please.
27-11-14 Ro LowSafeMove2 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 602 W: 88 L: 197 D: 317
sprt @ 15+0.05 th 1 Full penalty whereas the previous run is half penalty
27-11-14 Ro LowSafeMove2 diff
LLR: -2.98 (-2.94,2.94) [-1.50,4.50]
Total: 1353 W: 231 L: 336 D: 786
sprt @ 15+0.05 th 1 Fixed bug, calculating mobility adjust in another loop.
26-11-14 jo Outposts2 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 19597 W: 3163 L: 3255 D: 13179
sprt @ 60+0.05 th 1 LTC: Check some handmade values. Final try.
26-11-14 Ro MobilityArea diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 4113 W: 762 L: 857 D: 2494
sprt @ 15+0.05 th 1 Extended MobilityArea, now include squares which contains opponent pieces (which were previously excluded because protected by pawn, etc)
26-11-14 Ro LowSafeMove2 diff
LLR: -0.00 (-2.94,2.94) [-1.50,4.50]
Total: 5 W: 1 L: 1 D: 3
sprt @ 15+0.05 th 1 Take 2: half penalties, and exclude cases where piece can exchange
24-11-14 jo Outposts2 diff
LLR: 2.95 (-2.94,2.94) [-0.50,3.50]
Total: 257343 W: 51883 L: 51012 D: 154448
sprt @ 15+0.05 th 1 STC: Check some handmade values. Final try.
26-11-14 Ro LowSafeMove2 diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 1845 W: 344 L: 447 D: 1054
sprt @ 15+0.05 th 1 Reduce mobility measure when no safe movements
25-11-14 na tune_futility diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 39957 W: 6771 L: 6796 D: 26390
sprt @ 60+0.05 th 1 LTC: Tuning attempt for futility. Use quadratic formula.
25-11-14 jo rook_outpost diff
LLR: -3.26 (-2.94,2.94) [-1.50,4.50]
Total: 7500 W: 1412 L: 1509 D: 4579
sprt @ 15+0.05 th 1 Rook outpost as proposed by Nimzowitsch.
25-11-14 jo Outposts diff
LLR: -3.41 (-2.94,2.94) [-1.50,4.50]
Total: 21867 W: 4301 L: 4364 D: 13202
sprt @ 15+0.05 th 1 Try to provoke a weakening pawn move.
25-11-14 ce kick_queen diff
LLR: -2.97 (-2.94,2.94) [0.00,6.00]
Total: 13249 W: 2160 L: 2200 D: 8889
sprt @ 60+0.05 th 1 Bonus for threatening an attacking queen