triton.experimental.gluon.language.distributed_type

class triton.experimental.gluon.language.distributed_type(self, element_ty: dtype, shape: List[int], layout)

Methods

`__init__`(self, element_ty, shape, layout)
`codegen_name`(self)
`get_block_shapes`(self)
`get_int_max_value`(self)
`get_int_min_value`(self)
`is_bf16`(self)
`is_block`(self)
`is_bool`(self)
`is_const`()
`is_dtype`(type_str)
`is_floating`(self)
`is_fp16`(self)
`is_fp32`(self)
`is_fp64`(self)
`is_fp8`(self)
`is_fp8e4b15`(self)
`is_fp8e4b8`(self)
`is_fp8e4nv`(self)
`is_fp8e5`(self)
`is_fp8e5b16`(self)
`is_int`(self)
`is_int1`(self)
`is_int16`(self)
`is_int32`(self)
`is_int64`(self)
`is_int8`(self)
`is_int_signed`(self)
`is_int_unsigned`(self)
`is_ptr`()
`is_standard_floating`(self)
`is_uint16`(self)
`is_uint32`(self)
`is_uint64`(self)
`is_uint8`(self)
`is_void`()
`kind`(self)
`mangle`(self)
`to_ir`(self, builder)
`with_element_ty`(self, scalar_ty)

Attributes

`FP_TYPES`
`OTHER_TYPES`
`SINT_TYPES`
`STANDARD_FP_TYPES`
`UINT_TYPES`
`cache_key_part`	See cache_key_part() in triton.cc.
`nbytes`
`scalar`