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 function

grouped_matmul_block_scaled_mxfp4_kernel

def grouped_matmul_block_scaled_mxfp4_kernel[out_dtype: DType, BLOCK_N: Int](c_ptr: Pointer[Scalar[out_dtype], MutAnyOrigin], a_ptr: Pointer[UInt8, ImmutAnyOrigin], b_ptr: Pointer[UInt8, ImmutAnyOrigin], a_scales_ptr: Pointer[Float8_e8m0fnu, ImmutAnyOrigin], b_scales_ptr: Pointer[Float8_e8m0fnu, ImmutAnyOrigin], row_offsets_ptr: Pointer[UInt32, ImmutAnyOrigin], expert_ids_ptr: Pointer[Int32, ImmutAnyOrigin], num_active_experts: Int32, M: Int32, N: Int32, K: Int32)