OpenBench Testing Framework

Active : 32 Machines / 64 Threads / 2.15 MNPS
Priority 0
MenkibLeela-dag-previewpath-repetition-informationdiff8.0+0.08LLR: 0.62 (-2.94, 2.94) [-3.00, 0.00]
Games: 57114 W: 14228 L: 14306 D: 28580
Ptnml(0-2): 4, 5724, 17182, 5640, 7
Add path dependant draw evaluations to repetitions
MenkibLeela-dag-previewfix-bound-update-thread-reorderingdiff8.0+0.08LLR: -0.31 (-2.94, 2.94) [0.00, 3.00]
Games: 3562 W: 859 L: 868 D: 1835
Ptnml(0-2): 0, 351, 1088, 342, 0
Increase error tolerances for value calculation
Priority -5
MenkibLeela-dag-previewtcec-tune-with-blocking-cuda *diff40.0+0.40Tuning 12 Parameters
58743/100000 Iterations
234972/400000 Games Played
Finished
MenkibLeela-dag-previewfetch-nn-results-taskdiff8.0+0.08LLR: 0.00 (-2.94, 2.94) [-3.00, 0.00]
Games: 0 W: 0 L: 0 D: 0
Ptnml(0-2): 0, 0, 0, 0, 0
Add tasks to fetch network results
MenkibLeela-dag-previewuci-output-taskdiff8.0+0.08LLR: 0.00 (-2.94, 2.94) [-3.00, 0.00]
Games: 0 W: 0 L: 0 D: 0
Ptnml(0-2): 0, 0, 0, 0, 0
Add a reset computation task
MenkibLeela-dag-previewfix-bound-update-thread-reordering-transposition-includeddiff8.0+0.08LLR: -1.71 (-2.94, 2.94) [-3.00, 0.00]
Games: 32588 W: 7985 L: 8136 D: 16467
Ptnml(0-2): 3, 3298, 9845, 3143, 5
Increase error tolerances for value calculation
MenkibLeela-dag-previewtcec-season-30-task-workersdiff8.0+0.08LLR: 2.95 (-2.94, 2.94) [-3.00, 0.00]
Games: 59990 W: 14852 L: 14815 D: 30323
Ptnml(0-2): 8, 5896, 18150, 5933, 8
Add missing fields to node to process debug output
MenkibLeela-dag-previewfix-bound-update-thread-reordering-pathless-repetitiondiff8.0+0.08LLR: 2.95 (-2.94, 2.94) [0.00, 3.00]
Games: 14128 W: 3585 L: 3407 D: 7136
Ptnml(0-2): 0, 1292, 4305, 1464, 3
Increase error tolerances for value calculation
MenkibLeela-dag-previewtcec-season-30-tt-owns-low-nodediff8.0+0.08LLR: 2.95 (-2.94, 2.94) [-3.00, 0.00]
Games: 40750 W: 10248 L: 10174 D: 20328
Ptnml(0-2): 9, 4021, 12237, 4103, 5
Store shared pointer directly in cache
MenkibLeela-dag-previewpath-repetition-information-transposition-includeddiff8.0+0.08LLR: -2.96 (-2.94, 2.94) [0.00, 3.00]
Games: 18714 W: 4665 L: 4781 D: 9268
Ptnml(0-2): 3, 1882, 5702, 1768, 2
Add path dependant draw evaluations to repetitions
MenkibLeela-dag-previewtcec-season-30-repetition-updatediff8.0+0.08LLR: 2.95 (-2.94, 2.94) [-3.00, 0.00]
Games: 14970 W: 3775 L: 3651 D: 7544
Ptnml(0-2): 4, 1426, 4501, 1550, 4
Trace scope contested spin mutex and variable spin limit.
MenkibLeela-dag-previewtcec-season-30-solid-edgesdiff8.0+0.08LLR: 2.95 (-2.94, 2.94) [-3.00, 0.00]
Games: 29888 W: 7463 L: 7372 D: 15053
Ptnml(0-2): 3, 2677, 9492, 2770, 2
Avoid losing tree in single child release
JohnspLeela-dag-previewt3d-kd2-03477500diffN=300Elo: -72.07 +- 5.44 (95%) [N=4000]
Games: 4000 W: 644 L: 1462 D: 1894
Ptnml(0-2): 41, 916, 866, 174, 3
Test T3d-KD2 to BT4-tf13tune at fixed 300 node count
MenkibLeela-dag-previewtcec-season-30-task-workersdiff8.0+0.08LLR: -0.15 (-2.94, 2.94) [0.00, 3.00]
Games: 36 W: 2 L: 19 D: 15
Ptnml(0-2): 2, 13, 3, 0, 0
Add uci output task.
MenkibLeela-dag-previewtcec-season-30-solid-edgesdiff8.0+0.08LLR: -1.17 (-2.94, 2.94) [-3.00, 0.00]
Games: 9304 W: 2254 L: 2331 D: 4719
Ptnml(0-2): 0, 936, 2859, 855, 2
Limit maximum allocation size of MakeSolidQueue.
KovaxLeela-dag-previewpolicy-blend-gpudiff10.0+0.10Elo: 20.21 +- 3.07 (95%) [N=10000]
Games: 10000 W: 2542 L: 1961 D: 5497
Ptnml(0-2): 1, 745, 2929, 1322, 3
no early stopping
KovaxLeela-dag-previewpolicy-blend-gpudiff10.0+0.10Tuning 3 Parameters
2500/2500 Iterations
40000/40000 Games Played
optimism blend tune, uci overwrites backendopts
MenkibLeela-dag-previewtcec-season-30-solid-edgesdiff8.0+0.08LLR: -0.46 (-2.94, 2.94) [-3.00, 0.00]
Games: 7784 W: 1885 L: 1923 D: 3976
Ptnml(0-2): 0, 773, 2384, 735, 0
Fix msvc build using aligned_malloc
KovaxLeela-dag-previewpolicy-blend-gpudiff10.0+0.10LLR: 2.94 (-2.94, 2.94) [0.00, 3.00]
Games: 2082 W: 568 L: 408 D: 1106
Ptnml(0-2): 0, 137, 609, 293, 2
testing policy blending on gpu with the params of cpu blending
MenkibLeela-dag-previewtcec-season-30-solid-edgesdiff8.0+0.08LLR: 2.95 (-2.94, 2.94) [-3.00, 0.00]
Games: 1234 W: 409 L: 225 D: 600
Ptnml(0-2): 0, 48, 337, 232, 0
Use waitable atomic for task worker state
MenkibLeela-dag-previewtcec-season-30-swissdiff1800.0+3.00LLR: 2.95 (-2.94, 2.94) [0.00, 3.00]
Games: 23838 W: 5924 L: 5745 D: 12169
Ptnml(0-2): 0, 1846, 8049, 2023, 1
Calibration Elo test for LCZeroCPU on TCEC. (100 Elo)
KovaxLeela-dag-previewdebl1M-k2hyb-100kdiff10.0+0.10LLR: 2.94 (-2.94, 2.94) [0.00, 3.00]
Games: 9740 W: 2267 L: 2088 D: 5385
Ptnml(0-2): 0, 971, 2753, 1142, 4
k2hyb recheck, as this net played vs 100k instead of 1M
MenkibLeela-dag-previewtcec-season-30-swissdiff1800.0+3.00LLR: -2.95 (-2.94, 2.94) [0.00, 3.00]
Games: 5404 W: 1275 L: 1406 D: 2723
Ptnml(0-2): 1, 519, 1792, 390, 0
Calibration Elo test for LCZeroCPU on TCEC. (200 Elo)
KovaxLeela-dag-previewpolicy-blenddiff10.0+0.10LLR: 2.96 (-2.94, 2.94) [0.00, 3.00]
Games: 8412 W: 1970 L: 1798 D: 4644
Ptnml(0-2): 1, 775, 2483, 945, 2
Optimistic blend vs Master post softmax.
KovaxLeela-dag-previewpolicy-blenddiff10.0+0.10Tuning 6 Parameters
3612/12500 Iterations
57794/200000 Games Played
Tune policy blending on T1 sharp net.
KovaxLeela-dag-previewmasterdiffN=1000Elo: 11.91 +- 1.76 (95%) [N=40000]
Games: 40014 W: 10593 L: 9222 D: 20199
Ptnml(0-2): 181, 3889, 10566, 5120, 251
check how much PST and FPU is shift alone gains with the sharp target corpus.
KovaxLeela-dag-previewpolicy-blenddiffN=1000Elo: 57.21 +- 2.44 (95%) [N=40000]
Games: 21086 W: 7049 L: 3608 D: 10429
Ptnml(0-2): 30, 1176, 4936, 4125, 276
small node count initial check on sharp policy target + policy head blending tune resutls
KovaxLeela-dag-previewpolicy-blenddiffN=1000Tuning 6 Parameters
3000/3000 Iterations
192016/192000 Games Played
Policy blend tune on fix nodes