Emulated FP32 GEMM for AMD CDNA (MI250/gfx90a) with an accuracy contract it actually keeps: 1.20-1.22x native FP32, full FP32 exponent range, and a cheap cancellation probe so it never silently under-delivers.
What is the Mazukiri/emugemm-cdna GitHub project? Description: "Emulated FP32 GEMM for AMD CDNA (MI250/gfx90a) with an accuracy contract it actually keeps: 1.20-1.22x native FP32, full FP32 exponent range, and a cheap cancellation probe so it never silently under-delivers.". Written in C++. Explain what it does, its main use cases, key features, and who would benefit from using it.
Question is copied to clipboard — paste it after the AI opens.
Clone via HTTPS
Clone via SSH
Download ZIP
Download main.zipReport bugs or request features on the emugemm-cdna issue tracker:
Open GitHub Issues