Dev Builds » 20261024-1808

You are viewing an old NCM Stockfish dev build test. You may find the most recent dev build tests using Stockfish 15 as the baseline here.

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 14. This yields an approximate Elo difference and establishes confidence in the strength of the dev builds.

Summary

Host Duration Avg Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN

Commit

Commit ID 7316ef4e86a7a4fcb1bfb909369dc35860a79e3d
Author Godsdar
Date 2026-10-24 18:08:59 UTC
Reject oversized Syzygy symbol tables `set_sizes()` reads the symbol count from the tablebase as a `u16`, but `btreeBuf`, `symlen` and `colour` have only `SymCount` (4096) entries. The existing `fits()` check bounds the source bytes, so a malformed table with a larger count can pass that check and overrun the destination in `memcpy()` and the subsequent symbol loop. Reject counts greater than `SymCount` through the existing corrupt-table path before copying or iterating. Counts up to and including `SymCount` keep their existing handling. This applies to both WDL and DTZ loading. Validation on macOS arm64 with Apple Clang: - Optimized Apple silicon build; changed translation unit compiled with `-Werror`. - All 75 upstream instrumented tests passed on a debug build with AddressSanitizer and UndefinedBehaviorSanitizer, including the Syzygy tests. - Baseline and patched default benches both searched 1,559,611 nodes. - `tests/signature.sh 1559611` and `git diff --check` passed. The bounds error was identified and checked by source inspection; a malformed-tablebase runtime reproducer was not run. Windows and Linux validation remain for CI. Prepared with OpenAI Codex, including source analysis and testing. No functional change for valid tablebases. closes https://github.com/official-stockfish/Stockfish/pull/7237 No functional change
Copyright 2011–2026 Next Chess Move LLC