215 lines
6.0 KiB
Python
215 lines
6.0 KiB
Python
|
# Copyright 2021 The JAX Authors.
|
||
|
#
|
||
|
# 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
|
||
|
#
|
||
|
# https://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.
|
||
|
|
||
|
"""Experimental IREE backend for JAX.
|
||
|
|
||
|
This backend is quite incomplete, but exists to allow experimenting with
|
||
|
using IREE to compile and run JAX computations instead of XLA.
|
||
|
"""
|
||
|
|
||
|
# pytype: skip-file
|
||
|
|
||
|
from __future__ import annotations
|
||
|
|
||
|
import os
|
||
|
import platform
|
||
|
from typing import Any, List, Sequence, Optional
|
||
|
|
||
|
import iree.compiler
|
||
|
import iree.runtime
|
||
|
|
||
|
from jax._src.config import flags
|
||
|
from jax._src.lib import xla_client
|
||
|
import numpy as np
|
||
|
|
||
|
FLAGS = flags.FLAGS
|
||
|
|
||
|
|
||
|
flags.DEFINE_string(
|
||
|
'jax_iree_backend', os.getenv('JAX_IREE_BACKEND', 'cpu'),
|
||
|
'IREE compiler backend to use.')
|
||
|
|
||
|
iree_compiler_map = {
|
||
|
"cpu" : "llvm-cpu",
|
||
|
"cuda" : "cuda",
|
||
|
"vmvx" : "vmvx",
|
||
|
"vulkan" : "vulkan-spirv"
|
||
|
}
|
||
|
|
||
|
iree_runtime_map = {
|
||
|
"cpu" : "local-task",
|
||
|
"cuda" : "cuda",
|
||
|
"vmvx" : "local-task",
|
||
|
"vulkan" : "vulkan"
|
||
|
}
|
||
|
|
||
|
class IreeDevice:
|
||
|
|
||
|
def __init__(self, client):
|
||
|
self.id = 0
|
||
|
self.host_id = 0
|
||
|
self.process_index = 0
|
||
|
self.platform = "iree"
|
||
|
self.device_kind = "IREE device"
|
||
|
self.client = client
|
||
|
|
||
|
def __str__(self) -> str:
|
||
|
return "IreeDevice"
|
||
|
|
||
|
def transfer_to_infeed(self, literal: Any):
|
||
|
raise NotImplementedError("transfer_to_infeed")
|
||
|
|
||
|
def transfer_from_outfeed(self, shape: xla_client.Shape):
|
||
|
raise NotImplementedError("transfer_to_outfeed")
|
||
|
|
||
|
def live_buffers(self) -> List[IreeBuffer]:
|
||
|
raise NotImplementedError("live_buffers")
|
||
|
|
||
|
|
||
|
class IreeBuffer:
|
||
|
|
||
|
def __init__(self, client, device, buffer):
|
||
|
self.client = client
|
||
|
self._device = device
|
||
|
assert device is not None
|
||
|
self._buffer = buffer
|
||
|
|
||
|
def copy_to_device(self, device):
|
||
|
return self
|
||
|
|
||
|
def __array__(self, dtype=None, context=None):
|
||
|
return np.asarray(self._buffer)
|
||
|
|
||
|
def to_iree(self):
|
||
|
return self._buffer
|
||
|
|
||
|
def platform(self):
|
||
|
return self.client.platform
|
||
|
|
||
|
def device(self):
|
||
|
return self._device
|
||
|
|
||
|
def block_until_ready(self) -> IreeBuffer:
|
||
|
return self # no async
|
||
|
|
||
|
# overrides repr on base class which expects _value and aval attributes
|
||
|
def __repr__(self): return f'IreeBuffer({np.asarray(self)})'
|
||
|
|
||
|
@property
|
||
|
def _value(self):
|
||
|
return np.asarray(self)
|
||
|
|
||
|
def copy_to_host_async(self):
|
||
|
return self
|
||
|
|
||
|
class IreeExecutable:
|
||
|
|
||
|
def __init__(self, client, devices, module_object, function_name):
|
||
|
self.client = client
|
||
|
self.traceback = None
|
||
|
self.fingerprint = None
|
||
|
self._devices = devices
|
||
|
self.module_object = module_object
|
||
|
self.function_name = function_name
|
||
|
|
||
|
def local_devices(self) -> List[IreeDevice]:
|
||
|
return self._devices
|
||
|
|
||
|
def execute(self, arguments: Sequence[IreeBuffer]) -> List[IreeBuffer]:
|
||
|
inputs = [arg.to_iree() for arg in arguments]
|
||
|
outputs = self.module_object[self.function_name](*inputs)
|
||
|
# TODO(phawkins): Have a way to just have it always return the list,
|
||
|
# regardless of arity.
|
||
|
if not isinstance(outputs, list) and not isinstance(outputs, tuple):
|
||
|
outputs = [outputs]
|
||
|
return [
|
||
|
IreeBuffer(self.client, self._devices[0], output) for output in outputs
|
||
|
]
|
||
|
|
||
|
|
||
|
class IreeClient:
|
||
|
|
||
|
def __init__(self,
|
||
|
*,
|
||
|
iree_backend: Optional[str] = None):
|
||
|
self.platform = "iree"
|
||
|
self.platform_version = "0.0.1"
|
||
|
self.runtime_type = "iree"
|
||
|
self.iree_backend = (FLAGS.jax_iree_backend if iree_backend is None
|
||
|
else iree_backend)
|
||
|
self.compiler_driver = iree_compiler_map[self.iree_backend]
|
||
|
self.runtime_driver = iree_runtime_map[self.iree_backend]
|
||
|
self.iree_config = iree.runtime.system_api.Config(self.runtime_driver)
|
||
|
self._devices = [IreeDevice(self)]
|
||
|
|
||
|
def process_index(self) -> int:
|
||
|
return 0
|
||
|
|
||
|
def device_count(self) -> int:
|
||
|
return len(self._devices)
|
||
|
|
||
|
def devices(self) -> List[IreeDevice]:
|
||
|
return self._devices
|
||
|
|
||
|
def local_devices(self) -> List[IreeDevice]:
|
||
|
return self._devices
|
||
|
|
||
|
def local_device_count(self) -> int:
|
||
|
return len(self._devices)
|
||
|
|
||
|
def get_default_device_assignment(
|
||
|
self,
|
||
|
num_replicas: int) -> List[IreeDevice]:
|
||
|
if num_replicas != 1:
|
||
|
raise NotImplementedError("Only single-device computations implemented")
|
||
|
return [self._devices[0]]
|
||
|
|
||
|
|
||
|
def compile(self, computation: str,
|
||
|
compile_options: xla_client.CompileOptions) -> IreeExecutable:
|
||
|
del compile_options # Ignored.
|
||
|
extra_args = []
|
||
|
# extra_args=["--mlir-print-ir-after-all"]
|
||
|
if platform.system() == "Darwin" and platform.machine() == "arm64":
|
||
|
extra_args += ["--iree-llvm-target-triple=arm64-apple-darwin21.5.0"]
|
||
|
iree_binary = iree.compiler.compile_str(
|
||
|
computation, target_backends=[self.compiler_driver], input_type="mhlo",
|
||
|
# extended_diagnostics=True,
|
||
|
extra_args=extra_args,
|
||
|
)
|
||
|
# Load it into the runtime.
|
||
|
vm_module = iree.runtime.VmModule.from_flatbuffer(
|
||
|
self.iree_config.vm_instance, iree_binary)
|
||
|
module_object = iree.runtime.load_vm_module(vm_module, self.iree_config)
|
||
|
return IreeExecutable(self, self._devices, module_object, "main")
|
||
|
|
||
|
def buffer_from_pyval(
|
||
|
self,
|
||
|
argument: Any,
|
||
|
device: Optional[IreeDevice],
|
||
|
force_copy: bool = True,
|
||
|
host_buffer_semantics: xla_client.HostBufferSemantics = xla_client
|
||
|
.HostBufferSemantics.ZERO_COPY
|
||
|
) -> IreeBuffer:
|
||
|
# TODO(phawkins): IREE's python API will accept a numpy array directly but
|
||
|
# may want to explicitly construct a lower level BufferView to avoid copies.
|
||
|
if device is None:
|
||
|
assert type(argument) is np.ndarray
|
||
|
device = self._devices[0]
|
||
|
return IreeBuffer(self, device, np.array(argument, copy=True))
|
||
|
|
||
|
|
||
|
def iree_client_factory():
|
||
|
return IreeClient()
|