For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).
Mojo struct
Struct_matmul_dynamic_block_scaled_amd
struct Struct_matmul_dynamic_block_scaled_amd[lane_bytes: Int = Int(16)]
Registers the matmul.dynamic.block.scaled.amd graph op with the graph compiler.
Parameters
- lane_bytes (
Int): Operand bytes per lane per MFMA — 16 for MXFP4 (default) or 32 for MXFP8. The kernel readsa/bas raw bytes, so this rather than the operand dtype selects the format, and with it the K extent (Kat MXFP8,K // 2at MXFP4).
Implemented traits
Methods
execute
static def execute(c: T, a: T, b: T, a_scales: T, b_scales: T, context: DeviceContext)