Stockfish Testing Queue

Pending - 1 tests 0.0 hrs

19-01-20 MJZ Blocked_posG5 diff
LLR: 0.59 (-2.94,2.94) [0.50,4.50]
Total: 1730 W: 236 L: 201 D: 1293
sprt @ 240+2.4 th 1 Blocked position, it seems that the problem come from poskey in Qsearch. Third try at higher depths (d > 28). Limited test at VLTC with low throughput (test needs high depth search).

Active - 2 tests

19-01-21 MJZ Blocked_posG6 diff
LLR: -0.05 (-2.94,2.94) [0.50,4.50]
Total: 474 W: 73 L: 74 D: 327
sprt @ 120+1.2 th 1 Blocked position. Try n°5 with extension for potentially blocked positions. Limited test at VLTC with low throughput (test needs high depth search).
19-01-21 MJZ Blocked_posG4 diff
LLR: -0.51 (-2.94,2.94) [0.50,4.50]
Total: 2817 W: 406 L: 422 D: 1989
sprt @ 120+1.2 th 1 Blocked position. Try n°4 at higher depths (d > 28) with less memory to avoid crashes and non_pawn_material condition removed. Limited test at VLTC with low throughput (test needs high depth search).

Finished - 887 tests

19-01-20 MJZ Blocked_posG4 diff
LLR: -0.66 (-2.94,2.94) [0.50,4.50]
Total: 3337 W: 534 L: 555 D: 2248
sprt @ 60+0.6 th 1 Blocked position, it seems that the problem come from poskey in Qsearch. Second try with poskey change if rule50 > 20. Limited test at LTC with low throughput (test needs high depth search).
19-01-19 MJZ Blocked_posG4 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 13846 W: 2219 L: 2356 D: 9271
sprt @ 60+0.6 th 1 Blocked position, it seems that the problem come from poskey in Qsearch. Good results in some static positions. Limited test at LTC with low throughput (test needs high depth search).
19-01-19 MJZ Impr-tests diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 12971 W: 2812 L: 2912 D: 7247
sprt @ 10+0.1 th 1 Some tests in "improving" indicator. true if ss >= ss-4
19-01-18 MJZ Quick_Eval2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 14000 W: 2972 L: 3067 D: 7961
sprt @ 10+0.1 th 1 Some tries to see if we can make a quicker eval in special positions. Skip initiative and space if eg && mg > 360.
19-01-18 MJZ Quick_Eval diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 6569 W: 1375 L: 1507 D: 3687
sprt @ 10+0.1 th 1 Some tries to see if we can make a quicker eval in special positions. Skip initiative and space if eg && mg > 400.
19-01-17 MJZ Quick_Eval diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 30889 W: 6763 L: 6775 D: 17351
sprt @ 10+0.1 th 1 Some tries to see if we can make a quicker eval in some positions. Skip initiative if eg && mg > 400.
19-01-16 MJZ SearchTree-EG diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 18360 W: 4009 L: 4083 D: 10268
sprt @ 10+0.1 th 1 Extension for unclear EG if in CST. rule50 <=12, depth >=3+ply/2, && improving
19-01-16 MJZ SearchTree-EG diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 9904 W: 2058 L: 2173 D: 5673
sprt @ 10+0.1 th 1 Extension for unclear EG if in CST. rule50 <=12, depth >=3+ply/2
19-01-14 MJZ Value_Reduction diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 17343 W: 3737 L: 3816 D: 9790
sprt @ 10+0.1 th 1 Increase reduction for almost winning/losing position. abs(staticEval) > 400 + NPM / 32
19-01-13 MJZ EG_reduction7 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 14746 W: 3248 L: 3340 D: 8158
sprt @ 10+0.1 th 1 EG reduction for almost winning positions. !pvHit, eval > 2 pawns, nbr pawns > 3
19-01-13 MJZ EG_reduction8 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 51713 W: 11463 L: 11372 D: 28878
sprt @ 10+0.1 th 1 EG reduction for almost winning positions. !pvHit, eval > 2 pawns, nbr pawns > 2
19-01-13 MJZ EG_reduction7 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 23780 W: 5169 L: 5216 D: 13395
sprt @ 10+0.1 th 1 EG reduction for almost winning positions. !pvHit, eval > 2 pawns, nbr pawns > 2, reduction += depth / 4
19-01-13 MJZ EG_reduction7 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 27524 W: 6070 L: 6098 D: 15356
sprt @ 10+0.1 th 1 EG reduction for almost winning positions. !pvHit, eval > 2 pawns, ply > 4
19-01-12 MJZ EG_reduction8 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 48869 W: 10681 L: 10605 D: 27583
sprt @ 10+0.1 th 1 EG reduction for almost winning positions. !pvHit, eval > 2 pawns, even capture or promotion
19-01-12 MJZ EG_reduction7 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 43009 W: 9452 L: 9404 D: 24153
sprt @ 10+0.1 th 1 EG reduction for almost winning positions. !pvHit, eval > 2 pawns
19-01-12 MJZ EG_reduction8 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 12485 W: 2693 L: 2796 D: 6996
sprt @ 10+0.1 th 1 EG reduction for almost winning positions. !pvHit, eval > 2 pawns, movecount > 6
19-01-12 MJZ Blocked_NM diff
ELO: -15.48 +-11.9 (95%) LOS: 0.5%
Total: 1213 W: 197 L: 251 D: 765
5000 @ 60+0.6 th 1 Blocked position - 5k games with low throughput for first checks - increase verification depth to 20
19-01-12 MJZ Blocked_NM diff
ELO: -53.28 +-22.2 (95%) LOS: 0.0%
Total: 368 W: 45 L: 101 D: 222
5000 @ 60+0.6 th 1 Blocked position - 5k games with low throughput for first check
19-01-11 MJZ Blocked_NM diff
ELO: -60.28 +-15.1 (95%) LOS: 0.0%
Total: 879 W: 116 L: 267 D: 496
5000 @ 60+0.6 th 1 Blocked position - 5k games for first check at LTC
19-01-10 MJZ SearchTree-C1 diff
LLR: -2.95 (-2.94,2.94) [0.00,3.50]
Total: 75064 W: 12468 L: 12434 D: 50162
sprt @ 60+0.6 th 1 SearchTree - skip early pruning (speculative LTC, STC is very slow)
19-01-11 MJZ CST_add_lines diff
ELO: -6.77 +-6.6 (95%) LOS: 2.2%
Total: 4622 W: 952 L: 1042 D: 2628
5000 @ 10+0.1 th 1 5k games to check sanity
19-01-10 MJZ CST_add_lines diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 6418 W: 1323 L: 1455 D: 3640
sprt @ 10+0.1 th 1 Critical Search Tree : add lines for nonPV nodes if value > alpha - add depth condition > 6
19-01-10 MJZ CST_add_lines diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 7996 W: 1739 L: 1864 D: 4393
sprt @ 10+0.1 th 1 Critical Search Tree : add lines for nonPV nodes if value > alpha - First try
19-01-09 MJZ SearchTree-C1 diff
LLR: 2.96 (-2.94,2.94) [0.50,4.50]
Total: 81330 W: 18002 L: 17439 D: 45889
sprt @ 10+0.1 th 1 SearchTree - skip early pruning
19-01-10 MJZ CST_MCP diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 26392 W: 5737 L: 5772 D: 14883
sprt @ 10+0.1 th 1 For critical search tree lines, add 2 moves before MCP if bestValue < alpha - new run after correction
19-01-10 MJZ CST_MCP diff
LLR: -1.46 (-2.94,2.94) [0.50,4.50]
Total: 2871 W: 588 L: 655 D: 1628
sprt @ 10+0.1 th 1 For critical search tree lines, add 2 moves before MCP if bestValue < alpha
19-01-10 MJZ CST_MCP diff
LLR: -1.46 (-2.94,2.94) [0.50,4.50]
Total: 16563 W: 3589 L: 3589 D: 9385
sprt @ 10+0.1 th 1 For critical search tree lines, avoid MCP until bestvalue >= alpha
19-01-09 MJZ SearchTree-EG diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 31946 W: 6972 L: 6979 D: 17995
sprt @ 10+0.1 th 1 Extension for unclear EG if pvHit
19-01-09 MJZ SearchTree-C2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 34960 W: 7577 L: 7570 D: 19813
sprt @ 10+0.1 th 1 SearchTree - skip early pruning + depth limit = 4 (probably ealry pruning is only effective at low depths)
19-01-09 MJZ SearchTree-C diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 23471 W: 5077 L: 5149 D: 13245
sprt @ 10+0.1 th 1 SearchTree - tune depth limit, 4 instead of 6
19-01-09 MJZ SearchTree-D2 diff
LLR: 0.53 (-2.94,2.94) [0.00,4.00]
Total: 24704 W: 5428 L: 5302 D: 13974
sprt @ 10+0.1 th 1 Trying some parameter tuning for last passed commit. depth limit = 3 instead of 6
19-01-09 MJZ SearchTree-D1 diff
LLR: 0.71 (-2.94,2.94) [0.00,4.00]
Total: 26271 W: 5813 L: 5670 D: 14788
sprt @ 10+0.1 th 1 Trying some parameter tuning for last passed commit. depth limit = 4 instead of 6
19-01-09 MJZ SearchTree-D2 diff
LLR: 0.20 (-2.94,2.94) [-3.00,1.00]
Total: 12883 W: 2785 L: 2799 D: 7299
sprt @ 10+0.1 th 1 Trying some parameter tuning for last passed commit. Remove depth limit.
19-01-08 MJZ SearchTree-D1 diff
LLR: -2.32 (-2.94,2.94) [0.00,4.00]
Total: 54500 W: 12033 L: 11947 D: 30520
sprt @ 10+0.1 th 1 Trying some parameter tuning for last passed commit. depth limit = 5 instead of 6
19-01-08 MJZ SearchTree-A diff
LLR: 2.95 (-2.94,2.94) [0.00,3.50]
Total: 52452 W: 8828 L: 8507 D: 35117
sprt @ 60+0.6 th 1 SearchTree - LTC : introducing new concept, save principal lines into TT to generate a "search tree". For instance we just reduce reduction for these lines and skip early pruning. But a lot of other ideas are possible.
19-01-08 MJZ SearchTree-B diff
LLR: 1.01 (-2.94,2.94) [0.50,4.50]
Total: 24391 W: 5427 L: 5251 D: 13713
sprt @ 10+0.1 th 1 SearchTree B : reduce reduction && extend endgame positions if pvHit
19-01-08 MJZ SearchTree-A diff
LLR: 3.18 (-2.94,2.94) [0.50,4.50]
Total: 25317 W: 5645 L: 5345 D: 14327
sprt @ 10+0.1 th 1 SearchTree - tests seems OK - reduce reduction if pvHit
19-01-08 MJZ SearchTree-A diff
LLR: -1.02 (-2.94,2.94) [0.50,4.50]
Total: 12324 W: 2733 L: 2729 D: 6862
sprt @ 10+0.1 th 1 SearchTree - test 10k games to check sanity after correction
19-01-07 MJZ SearchTree-A diff
LLR: 0.42 (-2.94,2.94) [0.50,4.50]
Total: 13672 W: 3060 L: 2969 D: 7643
sprt @ 10+0.1 th 1 SearchTree - first test 10k games to check sanity
19-01-07 MJZ PV_PosA6 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 4286 W: 865 L: 1008 D: 2413
sprt @ 10+0.1 th 1 Search tree - random extension if pvHit after move, all branchs alpha > 0
19-01-07 MJZ PV_PosA7 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 2712 W: 514 L: 664 D: 1534
sprt @ 10+0.1 th 1 Search tree - random extension if pvHit after move, only PvNodes
19-01-06 MJZ PV_PosA4 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 28282 W: 6191 L: 6216 D: 15875
sprt @ 10+0.1 th 1 Save search PV positions. Revert last + for nonPV nodes depth > 7
19-01-06 MJZ PV_PosA4 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 4553 W: 926 L: 1068 D: 2559
sprt @ 10+0.1 th 1 Save search PV positions. Do opposite : increase reduction for non pvHit nodes & depth > 8
19-01-06 MJZ PV_PosA4 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 29655 W: 6440 L: 6459 D: 16756
sprt @ 10+0.1 th 1 Save search PV positions. Add some nonPV nodes in search tree if pvHit && value > alpha + remove rootNode
19-01-04 MJZ PV_PosA4 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 12373 W: 2622 L: 2725 D: 7026
sprt @ 10+0.1 th 1 Save search PV positions. Add some nonPV nodes in search tree if pvHit && value > alpha
19-01-04 MJZ PV_PosA2 diff
LLR: -1.90 (-2.94,2.94) [0.50,4.50]
Total: 31235 W: 6769 L: 6722 D: 17744
sprt @ 10+0.1 th 1 Save search PV positions. Reduce reduction if pvHit && movecount < 10
19-01-04 MJZ PV_PosA diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 36797 W: 7978 L: 7962 D: 20857
sprt @ 10+0.1 th 1 Extend last nodes of search tree
19-01-04 MJZ PV_PosA2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 18112 W: 3921 L: 3996 D: 10195
sprt @ 10+0.1 th 1 Save search PV positions. Reduce reduction x2 if pvHit
19-01-03 MJZ PV_PosA2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 69048 W: 15199 L: 15023 D: 38826
sprt @ 10+0.1 th 1 Save search PV positions. Reduce reduction if pvHit - lower depth limit to 6
19-01-03 MJZ PV_PosA diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 21127 W: 4547 L: 4608 D: 11972
sprt @ 10+0.1 th 1 Save search PV positions. Reduce reduction if pvHit - depth limit = 5