triton.language.tensor_descriptor

class triton.language.tensor_descriptor(self, handle, shape: List[tensor], strides: List[tensor], block_type: block_type)

A descriptor representing a tensor in global memory.

__init__(self, handle, shape: List[tensor], strides: List[tensor], block_type: block_type)

Not called by user code.

Methods

__init__(self, handle, shape, strides, ...)

Not called by user code.

atomic_add(self, offsets, value)

atomic_and(self, offsets, value)

atomic_max(self, offsets, value)

atomic_min(self, offsets, value)

atomic_or(self, offsets, value)

atomic_xor(self, offsets, value)

gather(self, *args)

Gather multiple descriptors worth of data

load(self, offsets)

Load a block from the descriptor starting at the given element offsets.

scatter(self, value, *args)

Scatter multiple descriptors worth of data

store(self, offsets, value)

Store a block from the descriptor starting at the given element offsets.

Attributes

block_shape

block_type

dtype

type