Luminex 5.12.0 - new version chess engine

Luminex - a UCI chess engine written in C++23. All evaluation values are self-engineered from chess first principles.

Author: Cheng Chang

Luminex 5.11.0 what's new?

Qsearch Speed Revolution

This release dramatically speeds up quiescence search — the engine's tactical computation core that runs at every leaf node. By removing expensive quiet check generation and replacing full sort with incremental pick-best, Luminex searches 13% faster at bullet time controls, translating to +57 Elo.

What's New

• Qsearch quiet check removal: Generating all ~30 quiet moves to find 2-3 checks was wasteful. Removing this gives +12.8% NPS with minimal tactical loss at bullet TC (1+0.01).
• Qsearch pick-best captures: Replaced std::sort with incremental pick-best selection for captures. When cutoff occurs on the first capture (common), the remaining captures are never sorted.
• Pinned enemy pieces eval: Bonus for enemy pieces pinned between their king and our sliders (Nimzowitsch).
• Capture check bonus: Pawn and knight captures that give check get +200k in PMG phase 2 ordering.
• King safety check ray hoist: Precompute king check rays outside per-piece loops for cleaner code.
• Compiler flags: Added -fno-exceptions -fno-rtti for GCC/Clang release builds.

Test Results

vs StashV19 (2473 Elo), tc=1+0.01, 200 games each:

Run	W-D-L	Cutechess Elo
Baseline (v5.11.0)	54-18-128	-135.0 ± 47.5
Qsearch checks removed	68-12-120	-92.5 ± 49.4
Confirmation run	71-11-118	-83.2 ± 48.4
+ Pick-best qsearch	72-12-116	-77.7 ± 48.1

Total: +57 Elo over v5.11.0 baseline, confirmed across 600 games.

Estimated Strength

~2395 Elo (v5.11.0 ~2338 + 57)

Binaries

Platform	File	Notes
Linux (modern)	luminex-linux-x86-64-modern	AVX2 + BMI2 (CPUs since ~2015)
Linux (compat)	luminex-linux-x86-64	SSE4.2 + POPCNT
Windows (modern)	luminex-windows-x86-64-modern.exe	ClangCL optimized
Windows (compat)	luminex-windows-x86-64.exe	MSVC compatible
macOS	luminex-macos-arm64	Apple Silicon

Luminex 5.12.0 download