File size: 8,028 Bytes
9c6594c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
import collections
from typing import Any, Union

import torch
from torch.types import Device

from . import _get_device_index, is_initialized


_device_t = Union[Device, str, int, None]


def empty_cache() -> None:
    r"""Release all unoccupied cached memory currently held by the caching
    allocator so that those can be used in other XPU application.

    .. note::
        :func:`~torch.xpu.empty_cache` doesn't increase the amount of XPU
        memory available for PyTorch. However, it may help reduce fragmentation
        of XPU memory in certain cases.
    """
    if is_initialized():
        torch._C._xpu_emptyCache()


def reset_peak_memory_stats(device: _device_t = None) -> None:
    r"""Reset the "peak" stats tracked by the XPU memory allocator.

    See :func:`~torch.xpu.memory_stats` for details. Peak stats correspond to the
    `"peak"` key in each individual stat dict.

    Args:
        device (torch.device or int or str, optional): selected device. Returns
            statistic for the current device, given by :func:`~torch.xpu.current_device`,
            if :attr:`device` is ``None`` (default).
    """
    device = _get_device_index(device, optional=True)
    return torch._C._xpu_resetPeakMemoryStats(device)


def reset_accumulated_memory_stats(device: _device_t = None) -> None:
    r"""Reset the "accumulated" (historical) stats tracked by the XPU memory allocator.

    See :func:`~torch.xpu.memory_stats` for details. Accumulated stats correspond to
    the `"allocated"` and `"freed"` keys in each individual stat dict.

    Args:
        device (torch.device or int or str, optional): selected device. Returns
            statistic for the current device, given by :func:`~torch.xpu.current_device`,
            if :attr:`device` is ``None`` (default).
    """
    device = _get_device_index(device, optional=True)
    return torch._C._xpu_resetAccumulatedMemoryStats(device)


def memory_stats_as_nested_dict(device: _device_t = None) -> dict[str, Any]:
    r"""Return the result of :func:`~torch.xpu.memory_stats` as a nested dictionary."""
    if not is_initialized():
        return {}
    device = _get_device_index(device, optional=True)
    return torch._C._xpu_memoryStats(device)


def memory_stats(device: _device_t = None) -> dict[str, Any]:
    r"""Return a dictionary of XPU memory allocator statistics for a given device.

    The return value of this function is a dictionary of statistics, each of
    which is a non-negative integer.

    Core statistics:

    - ``"allocated_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
      amount of allocated memory.
    - ``"reserved_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
      amount of reserved memory.
    - ``"active_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
      amount of active memory.
    - ``"requested_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
      memory requested by client code, compare this with allocated_bytes to check if
      allocation rounding adds too much overhead.

    For these core statistics, values are broken down as follows.

    Pool type:

    - ``all``: combined statistics across all memory pools.
    - ``large_pool``: statistics for the large allocation pool (for size >= 1MB allocations).
    - ``small_pool``: statistics for the small allocation pool (for size < 1MB allocations).

    Metric type:

    - ``current``: current value of this metric.
    - ``peak``: maximum value of this metric.
    - ``allocated``: historical total increase in this metric.
    - ``freed``: historical total decrease in this metric.

    Args:
        device (torch.device or int or str, optional): selected device. Returns
            statistics for the current device, given by :func:`~torch.xpu.current_device`,
            if :attr:`device` is ``None`` (default).
    """
    result = []

    def _recurse_add_to_result(prefix: str, obj: Any) -> None:
        if isinstance(obj, dict):
            if len(prefix) > 0:
                prefix += "."
            for k, v in obj.items():
                _recurse_add_to_result(prefix + k, v)
        else:
            result.append((prefix, obj))

    stats = memory_stats_as_nested_dict(device=device)
    _recurse_add_to_result("", stats)
    result.sort()

    return collections.OrderedDict(result)


def memory_allocated(device: _device_t = None) -> int:
    r"""Return the current GPU memory occupied by tensors in bytes for a given device.

    Args:
        device (torch.device or int or str, optional): selected device. Returns
            statistic for the current device, given by :func:`~torch.xpu.current_device`,
            if :attr:`device` is ``None`` (default).

    .. note::
        This is likely less than the amount shown in `xpu-smi` since some
        unused memory can be held by the caching allocator and some context
        needs to be created on GPU.
    """
    return memory_stats(device=device).get("allocated_bytes.all.current", 0)


def max_memory_allocated(device: _device_t = None) -> int:
    r"""Return the maximum GPU memory occupied by tensors in bytes for a given device.

    By default, this returns the peak allocated memory since the beginning of
    this program. :func:`~torch.xpu.reset_peak_memory_stats` can be used to
    reset the starting point in tracking this metric. For example, these two
    functions can measure the peak allocated memory usage of each iteration in a
    training loop.

    Args:
        device (torch.device or int or str, optional): selected device. Returns
            statistic for the current device, given by :func:`~torch.xpu.current_device`,
            if :attr:`device` is ``None`` (default).
    """
    return memory_stats(device=device).get("allocated_bytes.all.peak", 0)


def memory_reserved(device: _device_t = None) -> int:
    r"""Return the current GPU memory managed by the caching allocator in bytes for a given device.

    Args:
        device (torch.device or int or str, optional): selected device. Returns
            statistic for the current device, given by :func:`~torch.xpu.current_device`,
            if :attr:`device` is ``None`` (default).
    """
    return memory_stats(device=device).get("reserved_bytes.all.current", 0)


def max_memory_reserved(device: _device_t = None) -> int:
    r"""Return the maximum GPU memory managed by the caching allocator in bytes for a given device.

    By default, this returns the peak cached memory since the beginning of this
    program. :func:`~torch.xpu.reset_peak_memory_stats` can be used to reset
    the starting point in tracking this metric. For example, these two functions
    can measure the peak cached memory amount of each iteration in a training
    loop.

    Args:
        device (torch.device or int or str, optional): selected device. Returns
            statistic for the current device, given by :func:`~torch.xpu.current_device`,
            if :attr:`device` is ``None`` (default).
    """
    return memory_stats(device=device).get("reserved_bytes.all.peak", 0)


def mem_get_info(device: _device_t = None) -> tuple[int, int]:
    r"""Return the global free and total GPU memory for a given device.

    Args:
        device (torch.device or int or str, optional): selected device. Returns
            statistic for the current device, given by :func:`~torch.xpu.current_device`,
            if :attr:`device` is ``None`` (default).

    Returns:
        int: the memory available on the device in units of bytes.
        int: the total memory on the device in units of bytes
    """
    device = _get_device_index(device, optional=True)
    return torch._C._xpu_getMemoryInfo(device)


__all__ = [
    "empty_cache",
    "max_memory_allocated",
    "max_memory_reserved",
    "mem_get_info",
    "memory_allocated",
    "memory_reserved",
    "memory_stats",
    "memory_stats_as_nested_dict",
    "reset_accumulated_memory_stats",
    "reset_peak_memory_stats",
]