For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Mojo function

quantize_dynamic_block_scaled_mxfp4

def quantize_dynamic_block_scaled_mxfp4[in_dtype: DType](output: TileTensor[DType.uint8, Storage=output.Storage, address_space=output.address_space, linear_idx_type=output.linear_idx_type], output_scales: TileTensor[DType.float8_e8m0fnu, Storage=output_scales.Storage, address_space=output_scales.address_space, linear_idx_type=output_scales.linear_idx_type], input: TileTensor[in_dtype, Storage=input.Storage, address_space=input.address_space, linear_idx_type=input.linear_idx_type], ctx: DeviceContext)

Launches the AMD CDNA4 MXFP4 quantization kernel over a flat input buffer.

Enqueues quantize_dynamic_block_scaled_mxfp4_kernel with a 512-thread block grid, emitting a trace event for profiling.

Parameters:

in_dtype (DType): Element type of the input activation tensor (inferred). Must be bfloat16.