Table of Contents
Shortcuts

torch.xpu 的源代码

```html 
r"""
此包引入了对XPU后端的支持，特别是针对Intel GPU的优化。

此包是延迟初始化的，因此您可以始终导入它，并使用
:func:`is_available()` 来确定您的系统是否支持XPU。
"""
import threading
import traceback
from functools import lru_cache
from typing import Any, Callable, Dict, List, Optional, Tuple, Union

import torch
import torch._C
from .. import device as _device
from .._utils import _dummy_type, _LazySeedTracker
from ._utils import _get_device_index
from .streams import Event, Stream

_initialized = False
_tls = threading.local()
_initialization_lock = threading.Lock()
_queued_calls: List[
    Tuple[Callable[[], None], List[str]]
] = []  # 在初始化之前不要调用这些
_is_in_bad_fork = getattr(torch._C, "_xpu_isInBadFork", lambda: False)
_device_t = Union[_device, str, int, None]
_lazy_seed_tracker = _LazySeedTracker()
default_generators: Tuple[torch._C.Generator] = ()  # 类型: 忽略[赋值]


def _is_compiled() -> bool:
    r"""如果编译时支持XPU，则返回true。"""
    return torch._C._has_xpu


if _is_compiled():
    _XpuDeviceProperties = torch._C._XpuDeviceProperties
    _exchange_device = torch._C._xpu_exchangeDevice
    _maybe_exchange_device = torch._C._xpu_maybeExchangeDevice
else:
    # 如果PyTorch编译时不带XPU支持，定义虚拟对象
    _XpuDeviceProperties = _dummy_type("_XpuDeviceProperties")  # 类型: 忽略[赋值, 杂项]

    def _exchange_device(device: int) -> int:
        raise NotImplementedError("PyTorch编译时不带XPU支持")

    def _maybe_exchange_device(device: int) -> int:
        raise NotImplementedError("PyTorch编译时不带XPU支持")



[docs]@lru_cache(maxsize=1)
def device_count() -> int:
    r"""返回可用的XPU设备数量。"""
    if not _is_compiled():
        return 0
    return torch._C._xpu_getDeviceCount()




[docs]def is_available() -> bool:
    r"""返回一个布尔值，指示XPU当前是否可用。"""
    # 此函数从不抛出异常。
    return device_count() > 0



def is_bf16_supported():
    r"""返回一个布尔值，指示当前XPU设备是否支持dtype bfloat16。"""
    return True



[docs]def is_initialized():
    r"""返回PyTorch的XPU状态是否已初始化。"""
    return _initialized and not _is_in_bad_fork()



def _lazy_call(callable, **kwargs):
    if is_initialized():
        callable()
    else:
        global _lazy_seed_tracker
        if kwargs.get("seed_all", False):
            _lazy_seed_tracker.queue_seed_all(callable, traceback.format_stack())
        elif kwargs.get("seed", False):
            _lazy_seed_tracker.queue_seed(callable, traceback.format_stack())
        else:
            # 不存储实际的traceback以避免内存循环
            _queued_calls.append((callable, traceback.format_stack()))



[docs]def init():
    r"""初始化PyTorch的XPU状态。
    这是一个关于延迟初始化的Python API，避免在首次访问XPU之前初始化XPU。
    如果XPU状态已经初始化，则不执行任何操作。
    """
    _lazy_init()



def _lazy_init():
    global _initialized, _queued_calls
    if is_initialized() or <span class="
优云智算