Standard compilers use fixed -O3 pipelines that miss workload-specific speedups due to the Phase-Ordering Problem. Autotune Doctor is an AI-guided CLI that discovers, validates, and mathematically proves custom LLVM pipelines outperforming -O3. Unlike research tools, it tests on real silicon, enforces bitwise correctness gates, validates with Welch's t-test (p<0.05), and exports drop-in CMake recipes and CI guards. Install via PyPI: pip install autotune-doctor
Hey Product Hunt! π I built Autotune Doctor because standard compilers like Clang -O3 use a rigid, one-size-fits-all pass sequence that leaves performance on the table due to the compiler phase-ordering problem.
Autotune uses AI heuristics and genetic search to find, validate with bitwise correctness gates, and statistically prove custom LLVM pass pipelines tailored to your exact hardware.
Itβs 100% open-source and available on PyPI (pip install autotune-doctor). Would love your feedback and thoughts!