Stockfish Testing Queue

Finished - 1777 tests

18-04-08 pro ps_knights2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 18139 W: 3703 L: 3747 D: 10689
sprt @ 10+0.1 th 1 only apply penalty if pawn width > 4.
18-04-08 pro ps_knights diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 14991 W: 3028 L: 3087 D: 8876
sprt @ 10+0.1 th 1 knight pawn rule. S(0,20). increased penalty.
18-04-08 pro ps_rams3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13050 W: 2602 L: 2670 D: 7778
sprt @ 10+0.1 th 1 small endgame bonus for blocked enemy pawns in our bishop's color.
18-04-06 pro ps_rams2_tune diff
19584/20000 iterations
39468/40000 games played
40000 @ 20+0.2 th 1 tune new bonus/penalty for pawn rams on/off bishop colors.
18-04-05 pro ps_doubled2_tune diff
48011/50000 iterations
96646/100000 games played
100000 @ 20+0.2 th 1 tune a new per file doubled rule. Prio: -1, low throughput.
18-04-05 pro ps_bishopspawns3_tune diff
18415/20000 iterations
37114/40000 games played
40000 @ 20+0.2 th 1 tune a new rule for pawns penalties/bonuses for bishops. Prio -1, low throughput.
18-04-05 pro ps_bishopspawns2 diff
LLR: -0.29 (-2.94,2.94) [0.00,5.00]
Total: 1031 W: 205 L: 213 D: 613
sprt @ 10+0.1 th 1 if there is only one enemy bishop, bonus for pawns NOT on that color.
18-04-05 pro ps_bishopspawns diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 3505 W: 679 L: 792 D: 2034
sprt @ 10+0.1 th 1 Penalty for pawns ahead of bishop and on samecolor squares, bonus for pawns behind the bishop and on samecolor squares. Untuned.
18-04-05 pro ps_castlesafety4 diff
LLR: -2.71 (-2.94,2.94) [0.00,5.00]
Total: 27712 W: 5639 L: 5628 D: 16445
sprt @ 10+0.1 th 1 test tuned castling pseudo safety offsets. (-19, -9). Not optimistic.
18-04-05 pro ps_castlesafety3 diff
18137/20000 iterations
36507/40000 games played
40000 @ 20+0.2 th 1 one more run. Tune king/queen castling safety offsets.
18-04-04 pro ps_castlesafety3 diff
19577/20000 iterations
39417/40000 games played
40000 @ 20+0.2 th 1 tune king and queen side castling pseudo offsets.
18-04-05 pro ps_lazy50 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 50264 W: 10434 L: 10406 D: 29424
sprt @ 10+0.1 th 1 try both sides of yellow: 120.
18-04-05 pro ps_lazy50 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 41572 W: 8576 L: 8581 D: 24415
sprt @ 10+0.1 th 1 try both sides of yellow version. 80.
18-04-04 pro ps_lazy50 diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 41827 W: 8645 L: 8578 D: 24604
sprt @ 10+0.1 th 1 last try: linear again, 100 + count / 100.
18-04-04 pro ps_lazy50 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 2878 W: 528 L: 643 D: 1707
sprt @ 10+0.1 th 1 try a quadratic version, little change at low count, but increases quadraticly after 30 or so.
18-04-04 pro ps_lazy50 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13784 W: 2816 L: 2880 D: 8088
sprt @ 10+0.1 th 1 increase lazy eval according to the 50 move limit count. Less scaling.
18-04-04 pro ps_lazy50 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 10102 W: 2071 L: 2153 D: 5878
sprt @ 10+0.1 th 1 increase lazy eval according to the 50 move limit count.
18-04-04 pro ps_castlesafety3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13892 W: 2799 L: 2863 D: 8230
sprt @ 10+0.1 th 1 a bit less bonus to ensure motivation to castle.
18-04-04 pro ps_castlesafety diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 9149 W: 1813 L: 1992 D: 5344
sprt @ 10+0.1 th 1 test the value of the pseudo castling safety bonus.
18-04-04 pro ps_castlesafety2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 15684 W: 3213 L: 3268 D: 9203
sprt @ 10+0.1 th 1 apply 1/2 bonus but do both sides. Some bonus, but still motivation to castle. Losing castling rights decreases safety.
18-04-04 pro ps_castlesafety2 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5153 W: 964 L: 1068 D: 3121
sprt @ 10+0.1 th 1 only apply the pseudo castling bonus partially. 1/2 bonus and only one side.
18-04-04 pro ps_rooktune2 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 9868 W: 1977 L: 2101 D: 5790
sprt @ 10+0.1 th 1 remove psqt for rooks, and retune other rook parms.
18-04-03 pro ps_rooktune diff
45248/50000 iterations
91212/100000 games played
100000 @ 20+0.2 th 1 remove psqt for ROOK, then tune rook parameters.
18-04-03 pro ps_shelter14 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 30764 W: 6280 L: 6326 D: 18158
sprt @ 10+0.1 th 1 Try this version at 0,4 (passed -3,1 quite handily).
18-04-03 pro ps_shelter14 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 13858 W: 2746 L: 2855 D: 8257
sprt @ 10+0.1 th 1 null out rank7 and tuned. Just check if this will pass 0,4.
18-04-03 pro ps_shelter14_tune diff
16804/20000 iterations
33862/40000 games played
40000 @ 20+0.2 th 1 removing RANK7 may pass SPRT[0,4]. Tune a bit more for that purpose. Prio: -1, low throughput.
18-04-03 pro ps_rookbehind2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13342 W: 2662 L: 2729 D: 7951
sprt @ 10+0.1 th 1 updated rule: bonus for rooks being behind ALL enemy pawns.
18-04-03 pro ps_shelter17 diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 107655 W: 16492 L: 16806 D: 74357
sprt @ 60+0.6 th 1 LTC: remove ranks 5, 6, and 7 from the pawn shelter calculations .
18-04-02 pro ps_shelter17 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 109338 W: 22549 L: 22594 D: 64195
sprt @ 10+0.1 th 1 forgot a value. STC. remove ranks 5, 6, and 7 from pawn shelter.
18-04-02 pro ps_shelter17 diff
LLR: 0.07 (-2.94,2.94) [-3.00,1.00]
Total: 942 W: 186 L: 184 D: 572
sprt @ 10+0.1 th 1 remove ranks 5, 6, and 7 from pawn shelter.
18-04-01 pro ps_shelter16 diff
48954/50000 iterations
98686/100000 games played
100000 @ 20+0.2 th 1 run#4, still converging, on empty framework only. tune to remove rank5 from pawn shelter. Prio: -1, low throughput.
18-04-01 pro ps_shelter16 diff
18896/20000 iterations
38157/40000 games played
40000 @ 20+0.2 th 1 run#3: tune to remove 5th rank of pawn shelter.
18-04-01 pro ps_shelter15 diff
LLR: -2.96 (-2.94,2.94) [-3.00,1.00]
Total: 27273 W: 3992 L: 4183 D: 19098
sprt @ 60+0.6 th 1 empty framework. Verify shelter progress if there is nothing else to do.
18-04-01 pro ps_shelter16 diff
19839/20000 iterations
40000/40000 games played
40000 @ 20+0.2 th 1 run#2: updated values. tune to remove rank5 from pawn shelter.
18-04-01 pro ps_shelter16 diff
19840/20000 iterations
40000/40000 games played
40000 @ 20+0.2 th 1 tune to remove rank5 from pawn shelter.
18-03-31 pro ps_shelter15 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 26138 W: 5361 L: 5249 D: 15528
sprt @ 10+0.1 th 1 try to remove rank6 from king pawn shelter.
18-03-31 pro ps_shelter15_tune diff
17866/20000 iterations
36010/40000 games played
40000 @ 20+0.2 th 1 run#3 (updated values): tune to remove rank6 from pawn shelter.
18-03-31 pro ps_centerattacks diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6175 W: 1241 L: 1341 D: 3593
sprt @ 10+0.1 th 1 Bishop and Knights get bonus for attacking any Center square.
18-03-30 pro ps_shelter15_tune diff
18253/20000 iterations
36780/40000 games played
40000 @ 20+0.2 th 1 run#2 (updated values): tune to remove 6th rank from pawn shelter.
18-03-30 pro ps_shelter14 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 39926 W: 6024 L: 5932 D: 27970
sprt @ 60+0.6 th 1 LTC: remove rank7 from pawn shelter.
18-03-30 pro ps_shelter15_tune diff
9986/20000 iterations
20136/40000 games played
40000 @ 20+0.2 th 1 tune for removal of rank6 from pawn shelter.
18-03-30 pro ps_shelter14 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 15969 W: 3299 L: 3168 D: 9502
sprt @ 10+0.1 th 1 Remove Rank 7 from pawn shelter.
18-03-30 pro ps_shelter14_tune diff
18461/20000 iterations
37169/40000 games played
40000 @ 20+0.2 th 1 run#3. still converging. tune to remove RANK7 from pawn shelter.
18-03-30 pro ps_shelter14_tune diff
17934/20000 iterations
36169/40000 games played
40000 @ 20+0.2 th 1 run#2: updated values, still converging. tune to prepare to remove Rank7 from pawn shelter.
18-03-30 pro ps_shelter14_tune diff
19842/20000 iterations
40000/40000 games played
40000 @ 20+0.2 th 1 bugfix: tune to prepare to remove Rank7 from pawn shelter.
18-03-30 pro ps_shelter14 diff
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 57309 W: 11607 L: 11553 D: 34149
sprt @ 10+0.1 th 1 different approach: Changing weakness to strength. Test for non-regression. For information only.
18-03-29 pro ps_tunedsorting diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 57374 W: 11758 L: 11704 D: 33912
sprt @ 10+0.1 th 1 test tuned sorting values.
18-03-28 pro ps_sorting_tune diff
9933/10000 iterations
20000/20000 games played
20000 @ 20+0.2 th 1 tune sorting. . add another variable.
18-03-29 pro ps_moresupport diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 967 W: 188 L: 321 D: 458
sprt @ 10+0.1 th 1 random support idea (probably stinks).
18-03-28 pro ps_moredoubled diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 35127 W: 7303 L: 7266 D: 20558
sprt @ 10+0.1 th 1 doubled is set if there is a pawn anywhere behind us, not just the square behind.