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 struct
Struct_mxfp4_preshuffle_b_5d
struct Struct_mxfp4_preshuffle_b_5d
Run the AMD CDNA4 MXFP4 B 5D preshuffle as a custom op.
Used to pre-bake weights into Shuffler[E].b_5d_grouped_layout (the
layout the mxfp4_grouped_matmul_amd_preb reader expects) without
paying the >1 h CPU-side numpy shuffle on every model load.
Implemented traits
AnyType,
ImplicitlyDestructible
Methods
execute
static def execute[target: StringSlice[StaticConstantOrigin]](output: ManagedTensorSlice[Output, static_spec=output.static_spec], input: ManagedTensorSlice[Input, static_spec=input.static_spec], context: DeviceContext)
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!