Skip to main content

Mojo struct

BlockwiseFP8TilePayload

struct BlockwiseFP8TilePayload[a_type: DType, b_type: DType, a_scales_type: DType, a_shape: IndexList[2], b_shape: IndexList[2], a_scales_shape: IndexList[2], num_pipeline_stages: Int]

Tile payload for blockwise FP8 matmul (A, B, A-scales tiles).

Unlike BlockScaledTilePayload, this only stores A-scales in SMEM. B-scales are read directly from global memory during the epilogue phase.

Fields

  • a_tiles (BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].ATileArray):
  • b_tiles (BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].BTileArray):
  • a_scales_tiles (BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].AScalesTileArray):

Implemented traits

AnyType, Copyable, ImplicitlyCopyable, ImplicitlyDestructible, Movable, RegisterPassable, TilePayload, TrivialRegisterPassable

comptime members

AScalesTile

comptime AScalesTile = BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].AScalesTileArray.Tile

AScalesTileArray

comptime AScalesTileArray = SMemTileArray2DRowMajor[a_scales_type, a_scales_shape[0], a_scales_shape[1], num_pipeline_stages]

ATile

comptime ATile = BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].ATileArray.Tile

ATileArray

comptime ATileArray = SMemTileArray2D[a_type, a_shape[0], a_shape[1], num_pipeline_stages]

BTile

comptime BTile = BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].BTileArray.Tile

BTileArray

comptime BTileArray = SMemTileArray2D[b_type, b_shape[0], b_shape[1], num_pipeline_stages]

Methods

__init__

__init__(a_tiles: SMemTileArray2D[a_type, a_shape[0], a_shape[1], num_pipeline_stages], b_tiles: SMemTileArray2D[b_type, b_shape[0], b_shape[1], num_pipeline_stages], a_scales_tiles: SMemTileArray2DRowMajor[a_scales_type, a_scales_shape[0], a_scales_shape[1], num_pipeline_stages]) -> Self

get_tile

get_tile[k_group_size: Int](self, stage: UInt32, k_idx: Int) -> Tuple[BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].ATile, BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].BTile, BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].AScalesTile]

Get A, B, A-scales tiles at the specified stage and k-group index.

Returns:

Tuple

get_a_tile

get_a_tile[k_group_size: Int](self, stage: UInt32, k_idx: Int) -> BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].ATile

Get A tile at the specified stage and k-group index.

Returns:

BlockwiseFP8TilePayload

get_b_tile

get_b_tile[k_group_size: Int](self, stage: UInt32, k_idx: Int) -> BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].BTile

Get B tile at the specified stage and k-group index.

Returns:

BlockwiseFP8TilePayload

get_a_scales_tile

get_a_scales_tile[k_group_size: Int](self, stage: UInt32, k_idx: Int) -> BlockwiseFP8TilePayload[a_type, b_type, a_scales_type, a_shape, b_shape, a_scales_shape, num_pipeline_stages].AScalesTile

Get A-scales tile at the specified stage and k-group index.

Returns:

BlockwiseFP8TilePayload

Was this page helpful?