File amd_asm_matmul.py

extra/gemm/amd_asm_matmul.py:None–None · view source on GitHub ↗

Source from the content-addressed store, hash-verified

1	# RDNA3 128x128 tiled GEMM kernel - DSL version
2	# Computes C = A @ B for NxN float32 matrices using 128x128 tiles
3	#
4	# Architecture: RDNA3 (gfx1100)

nothing calls this directly

getenvFunction · 0.90

derive_fmac_patternFunction · 0.85

derive_permute_swapsFunction · 0.85

test_matmulFunction · 0.70

no test coverage detected