Enum BlockApi 

pub enum BlockApi {
    Unpacker,
    Systolic,
    Fragments,
}

How a matrix engine’s sub-block API runs one compute op over an osh × osw sub-block and its ibw inner steps: the calls it makes and the operand tiles it loads.

Variants

Unpacker

One call per inner step. The unpacker holds one operand tile along the sub-block’s shorter side and re-loads the other operand’s tiles for each: min(osh, osw) + osh · osw tiles a step. Tensix’s matmul_block (llk_unpack_AB_matmul: reuse_a = ct_dim >= rt_dim, t_dim held tiles, each running the other operand’s row of the block).

Systolic

One call over the whole inner block, its A and B panels loaded once: osh · ibw + ibw · osw tiles. The fabric’s systolic array, which pays its fill once per call.

Fragments

One fragment step per inner step. Each lane of the accumulator grid (SpatialTarget::accumulator_lane_grid) loads its own A rows and B columns: wn · osh + wm · osw tiles a step on a wm × wn grid, and two per output tile where no grid splits the sub-block evenly. The GPU’s wmma loop.

Trait Implementations

impl Clone for BlockApi

fn clone(&self) -> BlockApi

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BlockApi

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

Formats the value using the given formatter.

impl Hash for BlockApi

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for BlockApi

fn eq(&self, other: &BlockApi) -> 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 Serialize for BlockApi

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl Copy for BlockApi

impl Eq for BlockApi

impl StructuralPartialEq for BlockApi