Struct TileConfig 

pub struct TileConfig {
    pub grid_x: usize,
    pub grid_y: usize,
    pub per_core_m: usize,
    pub per_core_n: usize,
    pub in0_block_w: usize,
    pub out_subblock_h: usize,
    pub out_subblock_w: usize,
}

A tiling / blocking config for the blocked matmul kernel — the knobs an autotuner searches, in 32×32 tiles. best_tile_config fills this from a cost model; the vendor hand-tunes the equivalent.

Fields

grid_x: usize

Site columns (each owns per_core_n output tile-columns).

grid_y: usize

Site rows (each owns per_core_m output tile-rows).

per_core_m: usize

Output tile-rows computed per site.

per_core_n: usize

Output tile-cols computed per site.

in0_block_w: usize

Inner block width — contraction tiles streamed per block (bigger ⇒ more reuse per DRAM read).

out_subblock_h: usize

Accumulator-resident output sub-block height.

out_subblock_w: usize

Accumulator-resident output sub-block width.

Implementations

impl TileConfig

pub fn dst_tiles(&self) -> usize

Output tiles held in the accumulator register file per acquire — must be ≤ TileLimits::max_acc_tiles.

pub fn num_cores(&self) -> usize

Sites used.

Trait Implementations

impl Clone for TileConfig

fn clone(&self) -> TileConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TileConfig

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

Formats the value using the given formatter.

impl PartialEq for TileConfig

fn eq(&self, other: &TileConfig) -> 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 TileConfig

impl Eq for TileConfig

impl StructuralPartialEq for TileConfig