Files
paddlepaddle--paddle/python/paddle/device/custom_device.py
T
2026-07-13 12:40:42 +08:00

604 lines
20 KiB
Python

# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
#
# 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.
from __future__ import annotations
from typing import TYPE_CHECKING, TypeAlias
import paddle
from paddle.base import core
from .custom_streams import ( # noqa: F401
Event,
Stream,
create_event,
create_stream,
)
if TYPE_CHECKING:
from paddle import CustomPlace
_CustomPlaceLike: TypeAlias = (
CustomPlace
| str # some string like "iluvatar_gpu" "metax_gpu:0", etc.
| int # some int like 0, 1, etc.
)
dev_types = core.get_all_custom_device_type()
dev_type = dev_types[0] if dev_types else None
if dev_type and not core.is_compiled_with_custom_device(dev_type):
raise Exception(
"No custom device available, please install paddle with custom device support"
)
if dev_type and dev_type in ['metax_gpu', 'iluvatar_gpu']:
from .gpgpu_backend import get_device_properties
else:
from .default_backend import get_device_properties
__all__ = [
'Stream',
'Event',
'device_count',
'get_device_properties',
'empty_cache',
'max_memory_allocated',
'max_memory_reserved',
'reset_max_memory_allocated',
'reset_max_memory_reserved',
'memory_allocated',
'memory_reserved',
'current_stream',
'synchronize',
]
def device_count(device_type: str | None = None) -> int:
'''
Return the number of custom devices available.
Args:
device_type (str, optional): The type of custom device (e.g., 'npu', 'mlu', etc.).
If None, returns the count of the first available custom device type.
Returns:
int: the number of custom devices available.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.device_count()
>>> paddle.device.device_count('npu')
'''
if device_type:
num = core.get_custom_device_count(device_type)
else:
num = core.get_custom_device_count(dev_type)
return num
def empty_cache() -> None:
'''
Releases idle cached memory held by the allocator so that those can be used in other GPU
application and visible in device-specific tools.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.empty_cache()
'''
core.device_empty_cache()
def max_memory_allocated(device: _CustomPlaceLike | None = None) -> int:
'''
Return the peak size of memory that is allocated to tensor of the given device.
Args:
device(_CustomPlaceLike, optional): Support input like 'npu:0', 'mlu', int, or CustomPlace.
If None, the device is the first available custom device with index 0.
Returns:
int: The peak size of memory that is allocated to tensor of the given device, in bytes.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.max_memory_allocated('npu:0')
>>> paddle.device.max_memory_allocated('npu')
>>> paddle.device.max_memory_allocated(0)
>>> paddle.device.max_memory_allocated(Paddle.CustomPlace('npu', 0))
'''
device_id = 0
if device is None:
device_id = 0
elif isinstance(device, str):
colon_idx = device.rfind(':')
if colon_idx == -1:
device_id = 0
else:
device_id_str = device[colon_idx + 1 :]
if not device_id_str.isdigit():
raise ValueError(
f"Invalid device ID '{device_id_str}'. "
f"After colon must be digits only. "
"Example: 'npu:0'"
)
device_id = int(device_id_str)
elif isinstance(device, int):
device_id = device
elif isinstance(device, core.CustomPlace):
device_id = device.get_device_id()
else:
raise ValueError(
f"The input: {device} is not expected. Because paddle.device."
"max_memory_allocated only support str, int or CustomPlace. "
"Please input appropriate device again! "
"Example: 'npu:0'"
)
return core.device_memory_stat_peak_value("Allocated", device_id)
def max_memory_reserved(device: _CustomPlaceLike | None = None) -> int:
'''
Return the peak size of memory that is held by the allocator of the given device.
Args:
device(_CustomPlaceLike, optional): Support input like 'npu:0', 'mlu', int, or CustomPlace.
If None, the device is the first available custom device with index 0.
Returns:
int: The peak size of memory that is held by the allocator of the given device, in bytes.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.max_memory_reserved('npu:0')
>>> paddle.device.max_memory_reserved('npu')
>>> paddle.device.max_memory_reserved(0)
>>> paddle.device.max_memory_reserved(Paddle.CustomPlace('npu', 0))
'''
device_id = 0
if device is None:
device_id = 0
elif isinstance(device, str):
colon_idx = device.rfind(':')
if colon_idx == -1:
device_id = 0
else:
device_id_str = device[colon_idx + 1 :]
if not device_id_str.isdigit():
raise ValueError(
f"Invalid device ID '{device_id_str}'. "
f"After colon must be digits only. "
"Example: 'npu:0'"
)
device_id = int(device_id_str)
elif isinstance(device, int):
device_id = device
elif isinstance(device, core.CustomPlace):
device_id = device.get_device_id()
else:
raise ValueError(
f"The input: {device} is not expected. Because paddle.device."
"max_memory_reserved only support str, int or CustomPlace. "
"Please input appropriate device again! "
"Example: 'npu:0'"
)
return core.device_memory_stat_peak_value("Reserved", device_id)
def reset_max_memory_allocated(device: _CustomPlaceLike | None = None) -> None:
'''
Reset the peak size of memory that is allocated to tensor of the given device.
Args:
device(_CustomPlaceLike, optional): Support input like 'npu:0', 'mlu', int, or CustomPlace.
If None, the device is the first available custom device with index 0.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.reset_max_memory_allocated('npu:0')
>>> paddle.device.reset_max_memory_allocated('npu')
>>> paddle.device.reset_max_memory_allocated(0)
>>> paddle.device.reset_max_memory_allocated(Paddle.CustomPlace('npu', 0))
'''
device_id = 0
if device is None:
device_id = 0
elif isinstance(device, str):
colon_idx = device.rfind(':')
if colon_idx == -1:
device_id = 0
else:
device_id_str = device[colon_idx + 1 :]
if not device_id_str.isdigit():
raise ValueError(
f"Invalid device ID '{device_id_str}'. "
f"After colon must be digits only. "
"Example: 'npu:0'"
)
device_id = int(device_id_str)
elif isinstance(device, int):
device_id = device
elif isinstance(device, core.CustomPlace):
device_id = device.get_device_id()
else:
raise ValueError(
f"The input: {device} is not expected. Because paddle.device."
"reset_max_memory_allocated only support str, int or CustomPlace. "
"Please input appropriate device again! "
"Example: 'npu:0'"
)
core.device_memory_stat_reset_peak_value("Allocated", device_id)
def reset_max_memory_reserved(device: _CustomPlaceLike | None = None) -> None:
'''
Reset the peak size of memory that is held by the allocator of the given device.
Args:
device(_CustomPlaceLike, optional): Support input like 'npu:0', 'mlu', int, or CustomPlace.
If None, the device is the first available custom device with index 0.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.reset_max_memory_reserved('npu:0')
>>> paddle.device.reset_max_memory_reserved('npu')
>>> paddle.device.reset_max_memory_reserved(0)
>>> paddle.device.reset_max_memory_reserved(Paddle.CustomPlace('npu', 0))
'''
device_id = 0
if device is None:
device_id = 0
elif isinstance(device, str):
colon_idx = device.rfind(':')
if colon_idx == -1:
device_id = 0
else:
device_id_str = device[colon_idx + 1 :]
if not device_id_str.isdigit():
raise ValueError(
f"Invalid device ID '{device_id_str}'. "
f"After colon must be digits only. "
"Example: 'npu:0'"
)
device_id = int(device_id_str)
elif isinstance(device, int):
device_id = device
elif isinstance(device, core.CustomPlace):
device_id = device.get_device_id()
else:
raise ValueError(
f"The input: {device} is not expected. Because paddle.device."
"reset_max_memory_reserved only support str, int or CustomPlace. "
"Please input appropriate device again! "
"Example: 'npu:0'"
)
core.device_memory_stat_reset_peak_value("Reserved", device_id)
def memory_allocated(device: _CustomPlaceLike | None = None) -> int:
'''
Return the current size of memory that is allocated to tensor of the given device.
Args:
device(_CustomPlaceLike, optional): Support input like 'npu:0', 'mlu', int, or CustomPlace.
If None, the device is the first available custom device with index 0.
Returns:
int: The current size of memory that is allocated to tensor of the given device, in bytes.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.memory_allocated('npu:0')
>>> paddle.device.memory_allocated('npu')
>>> paddle.device.memory_allocated(0)
>>> paddle.device.memory_allocated(Paddle.CustomPlace('npu', 0))
'''
device_id = 0
if device is None:
device_id = 0
elif isinstance(device, str):
colon_idx = device.rfind(':')
if colon_idx == -1:
device_id = 0
else:
device_id_str = device[colon_idx + 1 :]
if not device_id_str.isdigit():
raise ValueError(
f"Invalid device ID '{device_id_str}'. "
f"After colon must be digits only. "
"Example: 'npu:0'"
)
device_id = int(device_id_str)
elif isinstance(device, int):
device_id = device
elif isinstance(device, core.CustomPlace):
device_id = device.get_device_id()
else:
raise ValueError(
f"The input: {device} is not expected. Because paddle.device."
"memory_allocated only support str, int or CustomPlace. "
"Please input appropriate device again! "
"Example: 'npu:0'"
)
return core.device_memory_stat_current_value("Allocated", device_id)
def memory_reserved(device: _CustomPlaceLike | None = None) -> int:
'''
Return the current size of memory that is held by the allocator of the given device.
Args:
device(_CustomPlaceLike, optional): Support input like 'npu:0', 'mlu', int, or CustomPlace.
If None, the device is the first available custom device with index 0.
Returns:
int: The current size of memory that is held by the allocator of the given device, in bytes.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.memory_reserved('npu:0')
>>> paddle.device.memory_reserved('npu')
>>> paddle.device.memory_reserved(0)
>>> paddle.device.memory_reserved(Paddle.CustomPlace('npu', 0))
'''
device_id = 0
if device is None:
device_id = 0
elif isinstance(device, str):
colon_idx = device.rfind(':')
if colon_idx == -1:
device_id = 0
else:
device_id_str = device[colon_idx + 1 :]
if not device_id_str.isdigit():
raise ValueError(
f"Invalid device ID '{device_id_str}'. "
f"After colon must be digits only. "
"Example: 'npu:0'"
)
device_id = int(device_id_str)
elif isinstance(device, int):
device_id = device
elif isinstance(device, core.CustomPlace):
device_id = device.get_device_id()
else:
raise ValueError(
f"The input: {device} is not expected. Because paddle.device."
"memory_reserved only support str, int or CustomPlace. "
"Please input appropriate device again! "
"Example: 'npu:0'"
)
return core.device_memory_stat_current_value("Reserved", device_id)
def current_stream(device: _CustomPlaceLike | None = None) -> core.CustomStream:
'''
Return the current stream by the device.
Args:
device(_CustomPlaceLike, optional): Support input like 'npu:0', 'mlu', int, or CustomPlace.
If None, the device is the first available custom device with index 0.
Returns:
Stream: The stream to the device.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.current_stream('npu:0')
>>> paddle.device.current_stream('npu')
>>> paddle.device.current_stream(0)
>>> paddle.device.current_stream(Paddle.CustomPlace('npu', 0))
'''
device_id = 0
if device is None:
device_id = 0
elif isinstance(device, str):
colon_idx = device.rfind(':')
if colon_idx == -1:
device_id = 0
else:
device_id_str = device[colon_idx + 1 :]
if not device_id_str.isdigit():
raise ValueError(
f"Invalid device ID '{device_id_str}'. "
f"After colon must be digits only. "
"Example: 'npu:0'"
)
device_id = int(device_id_str)
elif isinstance(device, int):
device_id = device
elif isinstance(device, core.CustomPlace):
device_id = device.get_device_id()
else:
raise ValueError(
f"The input: {device} is not expected. Because paddle.device."
"current_stream only support str, int or CustomPlace. "
"Please input appropriate device again! "
"Example: 'npu:0'"
)
return core._get_current_custom_device_stream(dev_type, device_id)
def synchronize(device: _CustomPlaceLike | None = None) -> None:
"""
Wait for the compute on the given device to finish.
Args:
device(_CustomPlaceLike, optional): Support input like 'npu:0', 'mlu', int, or CustomPlace.
If None, the device is the first available custom device with index 0.
Examples:
.. code-block:: pycon
>>> import paddle
>>> paddle.device.synchronize('npu:0')
>>> paddle.device.synchronize('npu')
>>> paddle.device.synchronize(0)
>>> paddle.device.synchronize(Paddle.CustomPlace('npu', 0))
"""
device_id = 0
if device is None:
device_id = 0
elif isinstance(device, str):
colon_idx = device.rfind(':')
if colon_idx == -1:
device_id = 0
else:
device_id_str = device[colon_idx + 1 :]
if not device_id_str.isdigit():
raise ValueError(
f"Invalid device ID '{device_id_str}'. "
f"After colon must be digits only. "
"Example: 'npu:0'"
)
device_id = int(device_id_str)
elif isinstance(device, int):
device_id = device
elif isinstance(device, core.CustomPlace):
device_id = device.get_device_id()
else:
raise ValueError(
f"The input: {device} is not expected. Because paddle.device."
"synchronize only support str, int or CustomPlace. "
"Please input appropriate device again! "
"Example: 'npu:0'"
)
core._synchronize_custom_device(dev_type, device_id)
def get_rng_state(
device: _CustomPlaceLike | None = None,
) -> core.GeneratorState:
r'''
Get the random state for the default generator.
Returns:
Tensor: The random state tensor.
Examples:
.. code-block:: pycon
>>> # doctest: +REQUIRES(env:CUSTOM_DEVICE)
>>> import paddle
>>> paddle.device.get_rng_state()
'''
place = paddle.device.device_to_place(device)
if isinstance(place, core.CPUPlace):
return core.default_cpu_generator().get_state()
return core.default_custom_device_generator(place).get_state()
def set_rng_state(
new_state: core.GeneratorState, device: _CustomPlaceLike | None = None
) -> None:
"""
Set the random number generator state of the specified device.
Args:
new_state (core.GeneratorState): The desired RNG state to set.
This should be a state object previously obtained from ``get_rng_state()``.
device (DeviceLike, optional): The device to set the RNG state for.
If not specified, uses the current default device (as returned by ``paddle.framework._current_expected_place_()``).
Can be a device object, integer device ID, or device string.
Returns:
None
Examples:
.. code-block:: pycon
>>> import paddle
>>> # Save RNG state
>>> state = paddle.device.get_rng_state()
>>> # Do some random operations
>>> x = paddle.randn([2, 3])
>>> # Restore RNG state
>>> paddle.device.set_rng_state(state)
"""
place = paddle.device.device_to_place(device)
if isinstance(place, core.CPUPlace):
core.default_cpu_generator().set_state(new_state)
else:
core.default_custom_device_generator(place).set_state(new_state)
def manual_seed(seed: int) -> None:
r"""Set the seed for generating random numbers for the current Device.
.. warning::
If you are working with a multi-Device model, this function is insufficient
to get determinism. To seed all Devices, use :func:`manual_seed_all`.
If current Device is CPU, this function will set the seed of the default CPU generator.
Sets the seed for global default generator, which manages the random number generation.
Args:
seed(int): The random seed to set.
Returns:
None
Examples:
.. code-block:: pycon
>>> # doctest: +REQUIRES(env:CUSTOM_DEVICE)
>>> import paddle
>>> paddle.device.manual_seed(102)
>>> # paddle.cuda.manual_seed(102) is equivalent to paddle.device.manual_seed(102)
"""
seed = int(seed)
place = paddle.framework._current_expected_place()
if isinstance(place, core.CPUPlace):
core.default_cpu_generator().manual_seed(seed)
else:
core.default_custom_device_generator(place).manual_seed(seed)