Enum AttnMat 

pub enum AttnMat {
    Bf16(DevicePtr),
    Q2K(DevicePtr),
    Q3K(DevicePtr),
}

2026-09-25: An attention projection on the device: bf16, or raw Q2_K blocks. Q3_K is refused by AttnV41::gemm. 2026-09-25: A resident [N, K] row-major projection from the GGUF path: expanded bf16, or the raw Q2_K / Q3_K blocks (WeightDtype::Q2K / Q3K) that the K-quant GEMV and MMQ kernels read.

Variants

Bf16(DevicePtr)

Q2K(DevicePtr)

Q3K(DevicePtr)

Implementations

impl ResidentMat

pub fn at_rows(self, rows: usize, k: usize) -> ResidentMat

2026-09-25: The sub-matrix starting rows rows in (each row k weights long).

Trait Implementations

impl Clone for ResidentMat

fn clone(&self) -> ResidentMat

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ResidentMat

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

Formats the value using the given formatter.

impl Copy for ResidentMat