diff options
| author | Scott Gasch <[email protected]> | 2026-10-24 09:07:03 -0700 |
|---|---|---|
| committer | Scott Gasch <[email protected]> | 2026-10-24 09:07:03 -0700 |
| commit | 5f901d737b17c5f99b6e456d6c0c7d5e3c093dc4 (patch) | |
| tree | 6601dadbd36c8f45991c5dd821a3dbb2059baafe /src/GNUmakefile | |
| parent | 69525eb2f42ea9cca8255f6cc3b85e6b1312fd05 (diff) | |
More aggressive move reductions; Wilson bound overrides static signal when available
SPRT playtest (2000 games, sd/depth matched): -535 =875 +590, score=0.514,
llr=+1.04 (floated around +1.0 the whole run, never went negative) toward
H1=+2.94. Not a confirmed win, but searches meaningfully deeper (avg depth
14.35 vs 13.50, EBF 3.131 vs 3.305) at no measurable strength cost -- worth
banking and continuing to accumulate games against.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Diffstat (limited to 'src/GNUmakefile')
| -rw-r--r-- | src/GNUmakefile | 13 |
1 files changed, 13 insertions, 0 deletions
diff --git a/src/GNUmakefile b/src/GNUmakefile index e710834..041ffbd 100644 --- a/src/GNUmakefile +++ b/src/GNUmakefile @@ -195,6 +195,19 @@ endif ifdef MOVE_ORDERING_EVIDENCE_COLLECTION PROFILE += -DMOVE_ORDERING_EVIDENCE_COLLECTION endif +# 2026-10-23, one-off measurement: does IS_CAPTURE(mvHash) (optionally +# && pi->eNodeType==NODE_CUT) meaningfully predict a leftover move's +# fail-high rate, beyond what the existing joint FH table's five axes +# already capture? Neither signal is currently baked into the table at +# all. Bucketed counters only (chess.h's u64HashCapTries/FH, [2][2] -- +# [hash move is a capture][node type guess is CUT]), printed by +# script.c's _PrintSuiteReport -- cheap enough to leave compiled in for +# a --cpus 1 collection run; not wired into split.c's MP merge, same +# convention as the other single-threaded-only calibration counters +# (u64TotalCMEvidenceTries et al) in that same report. +ifdef HASHCAP_COLLECTION +PROFILE += -DHASHCAP_COLLECTION +endif ifdef BOUNDS_CHECKING PROFILE += -fbounds-checking endif |