# Copyright (c) Meta Platforms, Inc. and affiliates.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
"""
Fused elementwise operator definition.
"""
from typing import Iterable, List
from aitemplate import backend
from aitemplate.backend import registry
from aitemplate.compiler.base import Operator
from aitemplate.compiler.ops.common.elementwise import elementwise
from aitemplate.compiler.tensor_accessor import TensorAccessor
# pylint: disable=C0301,C0103,W0223
def _check_shapes_eq(shapes1, shapes2) -> bool:
if len(shapes1) != len(shapes2):
return False
for shape1, shape2 in zip(shapes1, shapes2):
if shape1 != shape2:
return False
return True
[docs]class fused_elementwise(Operator):
"""fused_elementwise operator is used internally.
It's the actual operator which does ++ codegen.
"""
def _check_output_shape(self) -> None:
outputs = self._attrs["outputs"]
shape = outputs[0]._attrs["shape"]
for i in range(1, len(outputs)):
if not _check_shapes_eq(shape, outputs[i]._attrs["shape"]):
raise RuntimeError(
"Output shapes of fused_elementwise Op do not match! Shape1: {}. Shape2: {}.".format(
shape, outputs[i]._attrs["shape"]
)
)
def _update_inputs_outputs(
self, inputs: Iterable[Operator], outputs: Iterable[Operator]
) -> None:
self._attrs["inputs"] = list(inputs)
self._attrs["input_accessors"] = [
TensorAccessor(tensor) for tensor in self._attrs["inputs"]
]
self._attrs["outputs"] = list(outputs)
self._attrs["output_accessors"] = [
TensorAccessor(output_tensor) for output_tensor in self._attrs["outputs"]
]
self._check_output_shape()
# Preserve original tensors in case there are scatter / gather fusions.
# Need to copy Tensor objects.
self._attrs["original_inputs"] = list(self._attrs["inputs"])
self._attrs["original_outputs"] = list(self._attrs["outputs"])
for tensor in inputs:
tensor._attrs["dst_ops"].add(self)
for tensor in outputs:
tensor._attrs["src_ops"].add(self)
def _check_constant(self) -> None:
if len(self._attrs["inputs"]) == 0:
raise RuntimeError(f"No inputs for fused_elementwise! {self}")
for input_tensor in self._attrs["inputs"]:
if not input_tensor.is_a_const_num():
return
raise NotImplementedError(
"Cannot handle the case that all inputs of a fused_elementwise are constant numbers! "
f"Please use Python to calculate directly. Operator: {self}"
)
def __init__(
self,
elementwise_ops: List[elementwise],
inputs: Iterable[Operator],
outputs: Iterable[Operator],
) -> None:
super().__init__()
if len(elementwise_ops) == 0:
raise RuntimeError(
"fused_elementwise argument elementwise_ops cannot be empty!"
)
# It is required that elementwise_ops need to be topologically sorted.
self._attrs["op"] = "fused_elementwise"
self._attrs["elementwise_ops"] = elementwise_ops
self._attrs["has_profiler"] = False
self._update_inputs_outputs(inputs, outputs)
self._set_depth()
self._check_constant()
def _get_op_attributes(self):
return {
"elementwise_ops": self._attrs["elementwise_ops"],
"inputs": self._attrs["inputs"],
"outputs": self._attrs["outputs"],
}
[docs] def gen_function(self) -> str:
target = backend.target.Target.current()
func_key = "{target}.{op}.gen_function".format(
target=target.name(), op=self._attrs["op"]
)
func = registry.get(func_key)
return func(self._attrs)
def _args_for_pseudo_code(self):
return [f"func={[op._attrs['func'] for op in self._attrs['elementwise_ops']]}"]