【2/N】add support wmma kernels for RDNA4(GFX1201)#250

Open

vivienfanghuagood wants to merge 3 commits intoROCm:mainfrom

vivienfanghuagood:fx1201-wmma-gemm

Commits on Mar 19, 2026

Add WMMA GEMM kernels and wave32 kernel adaptations for RDNA4
vivienfanghuagood
committed
Add LDS-buffered FP8 GEMM kernel for large-M compute-bound shapes
vivienfanghuagood
committed

Commits on Mar 20, 2026

Fix wave_reduce: replace while loop with range_constexpr
vivienfanghuagood
committed