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_ep_fused_silu_mxfp8
struct Struct_ep_fused_silu_mxfp8
Registers the ep.fused_silu.mxfp8 graph op with the graph compiler.
Implemented traits
Methods
execute
static def execute[fp8_dtype: DType, scales_dtype: DType, input_dtype: DType, target: StringSpan[ImmStaticOrigin], *, fuse_a_scale_preshuffle: Bool = False, max_padded_M: Int = Int(0), clamp_activation: Bool = False](output: ManagedTensorSlice[IOSpec[_, _].Output, static_spec=output.static_spec], scales: ManagedTensorSlice[IOSpec[_, _].Output, static_spec=scales.static_spec], input: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=input.static_spec], row_offsets: ManagedTensorSlice[IOSpec[_, _].Input, static_spec=row_offsets.static_spec], alpha: Float32, limit: Float32, context: DeviceContext)
Execute the EP fused SILU kernel with MXFP8 quantization.
Same body as ep.fused_silu.mxfp4: fused_silu_mx_kernel takes its
element packing from the output dtype, so one fp8_e4m3fn byte per
element here rather than two FP4 nibbles, leaving output at the full
hidden size along axis 1. With fuse_a_scale_preshuffle, scales must
be shaped [n_local_experts * max_padded_M, K_SCALES].