triton.experimental.gluon.language.BlockedLayout

class triton.experimental.gluon.language.BlockedLayout(self, size_per_thread: ~typing.List[int], threads_per_warp: ~typing.List[int], warps_per_cta: ~typing.List[int], order: ~typing.List[int], cga_layout: ~typing.List[~typing.List[int]] = <factory>)

Represents a blocked layout, partitioning a tensor across threads, warps, and CTAs.

Parameters:

  • size_per_thread (List[int]) – Number of elements per thread per dimension.

  • threads_per_warp (List[int]) – Number of threads per warp per dimension.

  • warps_per_cta (List[int]) – Number of warps per CTA per dimension.

  • order (List[int]) – The ordering of dimensions for partitioning.

  • cga_layout (Optional[List[List[int]]]) – Bases describing how CTAs tile each dimension.

__init__(self, size_per_thread: ~typing.List[int], threads_per_warp: ~typing.List[int], warps_per_cta: ~typing.List[int], order: ~typing.List[int], cga_layout: ~typing.List[~typing.List[int]] = <factory>) → None

Methods

__init__(self, size_per_thread, ...)

format_hardware_view(self, shape)

format_tensor_view(self, shape)

mangle(self)

Attributes

rank

type

size_per_thread

threads_per_warp

warps_per_cta

order

cga_layout