Struct Fp8GroupedW8a8Layout 

pub struct Fp8GroupedW8a8Layout {
    pub act_s: usize,
    pub xq: usize,
    pub xs: usize,
    pub sh_s: usize,
}

2026-09-28: Byte offsets of the W8A8 activations inside the grouped decode’s buffers. The routed SiLU buffer (expert_gate_out, sized for [te, inter] FP32) holds the routed E4M3 products [te, inter], their scales [te, inter / 128] FP32, the quantized layer input [m, hidden] E4M3 and its scales [m, hidden / 128]; the shared SiLU buffer (logits, [m, inter] FP32) holds the shared products and their scales.

Fields

act_s: usize

xq: usize

xs: usize

sh_s: usize

Implementations

impl Fp8GroupedW8a8Layout

pub fn new( m: usize, top_k: usize, hidden: usize, inter: usize, routed_bytes: usize, shared_bytes: usize, ) -> Result<Self>

2026-09-28: The layout for m rows, or an error when hidden or inter is not whole 128 groups or either buffer is too small.

Trait Implementations

impl Clone for Fp8GroupedW8a8Layout

fn clone(&self) -> Fp8GroupedW8a8Layout

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Fp8GroupedW8a8Layout

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for Fp8GroupedW8a8Layout

fn eq(&self, other: &Fp8GroupedW8a8Layout) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for Fp8GroupedW8a8Layout

impl Eq for Fp8GroupedW8a8Layout

impl StructuralPartialEq for Fp8GroupedW8a8Layout