triton.language.tensor_descriptor

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

A descriptor representing a tensor in global memory.

__init__(handle, shape: List[tensor], strides: List[tensor], block_type: block_type): Not called by user code.

Methods

`__init__`(handle, shape, strides, block_type)	Not called by user code.
`atomic_add`(offsets, value[, _semantic])
`atomic_and`(offsets, value[, _semantic])
`atomic_max`(offsets, value[, _semantic])
`atomic_min`(offsets, value[, _semantic])
`atomic_or`(offsets, value[, _semantic])
`atomic_xor`(offsets, value[, _semantic])
`gather`(*args[, _semantic])	Gather multiple descriptors worth of data
`load`(offsets[, _semantic])	Load a block from the descriptor starting at the given element offsets.
`scatter`(value, *args[, _semantic])	Scatter multiple descriptors worth of data
`store`(offsets, value[, _semantic])	Store a block from the descriptor starting at the given element offsets.

Attributes