File size: 35,961 Bytes
c61ccee
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
r"""This package adds support for device memory management implemented in CUDA."""

import collections
import contextlib
import ctypes
import pickle
import sys
import warnings
from inspect import signature

from typing import Any, Dict, Optional, Tuple, Union

import torch
from torch import _C

from torch.types import Device
from .._utils import _dummy_type
from . import _get_device_index, _get_nvml_device_index, _lazy_init, is_initialized

from ._memory_viz import memory as _memory, segments as _segments

__all__ = [
    "caching_allocator_alloc",
    "caching_allocator_delete",
    "set_per_process_memory_fraction",
    "empty_cache",
    "memory_stats",
    "memory_stats_as_nested_dict",
    "reset_accumulated_memory_stats",
    "reset_peak_memory_stats",
    "reset_max_memory_allocated",
    "reset_max_memory_cached",
    "memory_allocated",
    "max_memory_allocated",
    "memory_reserved",
    "max_memory_reserved",
    "memory_cached",
    "max_memory_cached",
    "memory_snapshot",
    "memory_summary",
    "list_gpu_processes",
    "mem_get_info",
    "get_allocator_backend",
    "CUDAPluggableAllocator",
    "change_current_allocator",
]


if not hasattr(torch._C, "_cuda_CUDAAllocator"):
    # Define dummy base classes
    torch._C.__dict__["_cuda_CUDAAllocator"] = _dummy_type("_cuda_CUDAAllocator")


def _host_allocator():
    _lazy_init()
    return torch._C._cuda_cudaHostAllocator()


@contextlib.contextmanager
def _free_mutex():
    torch._C._cuda_lock_mutex()
    try:
        yield
    finally:
        torch._C._cuda_unlock_mutex()


def caching_allocator_alloc(size, device: Union[Device, int] = None, stream=None):
    r"""Perform a memory allocation using the CUDA memory allocator.



    Memory is allocated for a given device and a stream, this

    function is intended to be used for interoperability with other

    frameworks. Allocated memory is released through

    :func:`~torch.cuda.caching_allocator_delete`.



    Args:

        size (int): number of bytes to be allocated.

        device (torch.device or int, optional): selected device. If it is

            ``None`` the default CUDA device is used.

        stream (torch.cuda.Stream or int, optional): selected stream. If is ``None`` then

            the default stream for the selected device is used.



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    if device is None:
        device = torch.cuda.current_device()
    device = _get_device_index(device)
    if stream is None:
        stream = torch.cuda.current_stream(device)
    if isinstance(stream, torch.cuda.streams.Stream):
        stream = stream.cuda_stream
    if not isinstance(stream, int):
        raise TypeError(
            "Invalid type for stream argument, must be "
            "`torch.cuda.Stream` or `int` representing a pointer "
            "to a existing stream"
        )
    with torch.cuda.device(device):
        return torch._C._cuda_cudaCachingAllocator_raw_alloc(size, stream)


def caching_allocator_delete(mem_ptr):
    r"""Delete memory allocated using the CUDA memory allocator.



    Memory allocated with :func:`~torch.cuda.caching_allocator_alloc`.

    is freed here. The associated device and stream are tracked inside

    the allocator.



    Args:

        mem_ptr (int): memory address to be freed by the allocator.



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    torch._C._cuda_cudaCachingAllocator_raw_delete(mem_ptr)


def set_per_process_memory_fraction(

    fraction, device: Union[Device, int] = None

) -> None:
    r"""Set memory fraction for a process.



    The fraction is used to limit an caching allocator to allocated memory on a CUDA device.

    The allowed value equals the total visible memory multiplied fraction.

    If trying to allocate more than the allowed value in a process, will raise an out of

    memory error in allocator.



    Args:

        fraction(float): Range: 0~1. Allowed memory equals total_memory * fraction.

        device (torch.device or int, optional): selected device. If it is

            ``None`` the default CUDA device is used.

    .. note::

        In general, the total available free memory is less than the total capacity.

    """
    _lazy_init()
    if device is None:
        device = torch.cuda.current_device()
    device = _get_device_index(device)
    if not isinstance(fraction, float):
        raise TypeError("Invalid type for fraction argument, must be `float`")
    if fraction < 0 or fraction > 1:
        raise ValueError(f"Invalid fraction value: {fraction}. Allowed range: 0~1")

    torch._C._cuda_setMemoryFraction(fraction, device)


def empty_cache() -> None:
    r"""Release all unoccupied cached memory currently held by the caching

    allocator so that those can be used in other GPU application and visible in

    `nvidia-smi`.



    .. note::

        :func:`~torch.cuda.empty_cache` doesn't increase the amount of GPU

        memory available for PyTorch. However, it may help reduce fragmentation

        of GPU memory in certain cases. See :ref:`cuda-memory-management` for

        more details about GPU memory management.

    """
    if is_initialized():
        torch._C._cuda_emptyCache()


def memory_stats(device: Union[Device, int] = None) -> Dict[str, Any]:
    r"""Return a dictionary of CUDA 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.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:

      number of allocation requests received by the memory allocator.

    - ``"allocated_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:

      amount of allocated memory.

    - ``"segment.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:

      number of reserved segments from ``cudaMalloc()``.

    - ``"reserved_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:

      amount of reserved memory.

    - ``"active.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:

      number of active memory blocks.

    - ``"active_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:

      amount of active memory.

    - ``"inactive_split.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:

      number of inactive, non-releasable memory blocks.

    - ``"inactive_split_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:

      amount of inactive, non-releasable memory.



    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

      (as of October 2019, for size >= 1MB allocations).

    - ``small_pool``: statistics for the small allocation pool

      (as of October 2019, 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.



    In addition to the core statistics, we also provide some simple event

    counters:



    - ``"num_alloc_retries"``: number of failed ``cudaMalloc`` calls that

      result in a cache flush and retry.

    - ``"num_ooms"``: number of out-of-memory errors thrown.



    The caching allocator can be configured via ENV to not split blocks larger than a

    defined size (see Memory Management section of the Cuda Semantics documentation).

    This helps avoid memory fragmentation but may have a performance

    penalty. Additional outputs to assist with tuning and evaluating impact:



    - ``"max_split_size"``: blocks above this size will not be split.

    - ``"oversize_allocations.{current,peak,allocated,freed}"``:

      number of over-size allocation requests received by the memory allocator.

    - ``"oversize_segments.{current,peak,allocated,freed}"``:

      number of over-size reserved segments from ``cudaMalloc()``.



    The caching allocator can be configured via ENV to round memory allocations in order

    to reduce fragmentation. Sometimes the overhead from rounding can be higher than

    the fragmentation it helps reduce. The following stat can be used to check if

    rounding adds too much overhead:



    - ``"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.



    Args:

        device (torch.device or int, optional): selected device. Returns

            statistics for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.



    .. note::

        With :ref:`backend:cudaMallocAsync<cuda-memory-envvars>`, some stats are not

        meaningful, and are always reported as zero.

    """
    result = []

    def _recurse_add_to_result(prefix, obj):
        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_stats_as_nested_dict(device: Union[Device, int] = None) -> Dict[str, Any]:
    r"""Return the result of :func:`~torch.cuda.memory_stats` as a nested dictionary."""
    if not is_initialized():
        return {}
    device = _get_device_index(device, optional=True)
    return torch._C._cuda_memoryStats(device)


def reset_accumulated_memory_stats(device: Union[Device, int] = None) -> None:
    r"""Reset the "accumulated" (historical) stats tracked by the CUDA memory allocator.



    See :func:`~torch.cuda.memory_stats` for details. Accumulated stats correspond to

    the `"allocated"` and `"freed"` keys in each individual stat dict, as well as

    `"num_alloc_retries"` and `"num_ooms"`.



    Args:

        device (torch.device or int, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    device = _get_device_index(device, optional=True)
    return torch._C._cuda_resetAccumulatedMemoryStats(device)


def reset_peak_memory_stats(device: Union[Device, int] = None) -> None:
    r"""Reset the "peak" stats tracked by the CUDA memory allocator.



    See :func:`~torch.cuda.memory_stats` for details. Peak stats correspond to the

    `"peak"` key in each individual stat dict.



    Args:

        device (torch.device or int, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    device = _get_device_index(device, optional=True)
    return torch._C._cuda_resetPeakMemoryStats(device)


def reset_max_memory_allocated(device: Union[Device, int] = None) -> None:
    r"""Reset the starting point in tracking maximum GPU memory occupied by tensors for a given device.



    See :func:`~torch.cuda.max_memory_allocated` for details.



    Args:

        device (torch.device or int, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. warning::

        This function now calls :func:`~torch.cuda.reset_peak_memory_stats`, which resets

        /all/ peak memory stats.



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    warnings.warn(
        "torch.cuda.reset_max_memory_allocated now calls torch.cuda.reset_peak_memory_stats, "
        "which resets /all/ peak memory stats.",
        FutureWarning,
    )
    return reset_peak_memory_stats(device=device)


def reset_max_memory_cached(device: Union[Device, int] = None) -> None:
    r"""Reset the starting point in tracking maximum GPU memory managed by the caching allocator for a given device.



    See :func:`~torch.cuda.max_memory_cached` for details.



    Args:

        device (torch.device or int, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. warning::

        This function now calls :func:`~torch.cuda.reset_peak_memory_stats`, which resets

        /all/ peak memory stats.



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    warnings.warn(
        "torch.cuda.reset_max_memory_cached now calls torch.cuda.reset_peak_memory_stats, "
        "which resets /all/ peak memory stats.",
        FutureWarning,
    )
    return reset_peak_memory_stats(device=device)


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



    Args:

        device (torch.device or int, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. note::

        This is likely less than the amount shown in `nvidia-smi` since some

        unused memory can be held by the caching allocator and some context

        needs to be created on GPU. See :ref:`cuda-memory-management` for more

        details about GPU memory management.

    """
    return memory_stats(device=device).get("allocated_bytes.all.current", 0)


def max_memory_allocated(device: Union[Device, int] = 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.cuda.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, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    return memory_stats(device=device).get("allocated_bytes.all.peak", 0)


def memory_reserved(device: Union[Device, int] = 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, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    return memory_stats(device=device).get("reserved_bytes.all.current", 0)


def max_memory_reserved(device: Union[Device, int] = 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.cuda.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, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    return memory_stats(device=device).get("reserved_bytes.all.peak", 0)


def memory_cached(device: Union[Device, int] = None) -> int:
    r"""Deprecated; see :func:`~torch.cuda.memory_reserved`."""
    warnings.warn(
        "torch.cuda.memory_cached has been renamed to torch.cuda.memory_reserved",
        FutureWarning,
    )
    return memory_reserved(device=device)


def max_memory_cached(device: Union[Device, int] = None) -> int:
    r"""Deprecated; see :func:`~torch.cuda.max_memory_reserved`."""
    warnings.warn(
        "torch.cuda.max_memory_cached has been renamed to torch.cuda.max_memory_reserved",
        FutureWarning,
    )
    return max_memory_reserved(device=device)


def memory_snapshot():
    r"""Return a snapshot of the CUDA memory allocator state across all devices.



    Interpreting the output of this function requires familiarity with the

    memory allocator internals.



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    return torch._C._cuda_memorySnapshot()["segments"]


def memory_summary(device: Union[Device, int] = None, abbreviated: bool = False) -> str:
    r"""Return a human-readable printout of the current memory allocator statistics for a given device.



    This can be useful to display periodically during training, or when

    handling out-of-memory exceptions.



    Args:

        device (torch.device or int, optional): selected device. Returns

            printout for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).

        abbreviated (bool, optional): whether to return an abbreviated summary

            (default: False).



    .. note::

        See :ref:`cuda-memory-management` for more details about GPU memory

        management.

    """
    device = _get_device_index(device, optional=True)
    stats = memory_stats(device=device)

    def _format_size(sz, pref_sz):
        prefixes = ["B  ", "KiB", "MiB", "GiB", "TiB", "PiB"]
        prefix = prefixes[0]
        for new_prefix in prefixes[1:]:
            if pref_sz < 768 * 1024:
                break
            prefix = new_prefix
            sz //= 1024
            pref_sz /= 1024
        return f"{sz:6d} {prefix}"

    def _format_count(cnt, pref_cnt):
        prefixes = [" ", "K", "M"]
        prefix = prefixes[0]
        for new_prefix in prefixes[1:]:
            if pref_cnt < 750 * 1000:
                break
            prefix = new_prefix
            cnt //= 1000
            pref_cnt /= 1000
        return f"{cnt:7d} {prefix} "

    metrics_to_display = [
        ("allocated_bytes", "Allocated memory", _format_size),
        ("active_bytes", "Active memory", _format_size),
        ("requested_bytes", "Requested memory", _format_size),
        ("reserved_bytes", "GPU reserved memory", _format_size),
        ("inactive_split_bytes", "Non-releasable memory", _format_size),
        ("allocation", "Allocations", _format_count),
        ("active", "Active allocs", _format_count),
        ("segment", "GPU reserved segments", _format_count),
        ("inactive_split", "Non-releasable allocs", _format_count),
    ]

    lines = []
    lines.append("=" * 75)
    lines.append(" {_:16} PyTorch CUDA memory summary, device ID {device:<17d} ")
    lines.append("-" * 75)
    lines.append(
        "  {_:9} CUDA OOMs: {num_ooms:<12d} | {_:6} cudaMalloc retries: {num_alloc_retries:<8d}  "
    )
    lines.append("=" * 75)
    lines.append(
        "        Metric         | Cur Usage  | Peak Usage | Tot Alloc  | Tot Freed  "
    )

    for metric_key, metric_name, formatter in metrics_to_display:
        lines.append("-" * 75)
        submetrics = [("all", metric_name)]
        if not abbreviated:
            submetrics.append(("large_pool", "      from large pool"))
            submetrics.append(("small_pool", "      from small pool"))

        current_prefval, peak_prefval, allocated_prefval, freed_prefval = (
            None,
            None,
            None,
            None,
        )

        for submetric_key, submetric_name in submetrics:
            prefix = metric_key + "." + submetric_key + "."

            current = stats[prefix + "current"]
            peak = stats[prefix + "peak"]
            allocated = stats[prefix + "allocated"]
            freed = stats[prefix + "freed"]

            if current_prefval is None:
                current_prefval = current
                peak_prefval = peak
                allocated_prefval = allocated
                freed_prefval = freed

            lines.append(
                " {:<21} | {} | {} | {} | {} ".format(
                    submetric_name,
                    formatter(current, current_prefval),
                    formatter(peak, peak_prefval),
                    formatter(allocated, allocated_prefval),
                    formatter(freed, freed_prefval),
                ),
            )

    metrics_to_display = [
        ("oversize_allocations", "Oversize allocations", _format_count),
        ("oversize_segments", "Oversize GPU segments", _format_count),
    ]

    for metric_key, metric_name, formatter in metrics_to_display:
        lines.append("-" * 75)

        prefix = metric_key + "."

        current = stats[prefix + "current"]
        peak = stats[prefix + "peak"]
        allocated = stats[prefix + "allocated"]
        freed = stats[prefix + "freed"]

        lines.append(
            " {:<21} | {} | {} | {} | {} ".format(
                metric_name,
                formatter(current, current),
                formatter(peak, peak),
                formatter(allocated, allocated),
                formatter(freed, freed),
            ),
        )

    lines.append("=" * 75)

    fmt_dict = {"_": "", "device": device}
    for k, v in stats.items():
        fmt_dict[k.replace(".", "-")] = v
    return "|" + "|\n|".join(lines).format(**fmt_dict) + "|\n"


def list_gpu_processes(device: Union[Device, int] = None) -> str:
    r"""Return a human-readable printout of the running processes and their GPU memory use for a given device.



    This can be useful to display periodically during training, or when

    handling out-of-memory exceptions.



    Args:

        device (torch.device or int, optional): selected device. Returns

            printout for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).

    """
    try:
        import pynvml  # type: ignore[import]
    except ModuleNotFoundError:
        return "pynvml module not found, please install pynvml"
    from pynvml import NVMLError_DriverNotLoaded

    try:
        pynvml.nvmlInit()
    except NVMLError_DriverNotLoaded:
        return "cuda driver can't be loaded, is cuda enabled?"
    device = _get_nvml_device_index(device)
    handle = pynvml.nvmlDeviceGetHandleByIndex(device)
    procs = pynvml.nvmlDeviceGetComputeRunningProcesses(handle)
    lines = []
    lines.append(f"GPU:{device}")
    if len(procs) == 0:
        lines.append("no processes are running")
    for p in procs:
        mem = p.usedGpuMemory / (1024 * 1024)
        lines.append(f"process {p.pid:>10d} uses {mem:>12.3f} MB GPU memory")
    return "\n".join(lines)


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



    Args:

        device (torch.device or int, optional): selected device. Returns

            statistic for the current device, given by :func:`~torch.cuda.current_device`,

            if :attr:`device` is ``None`` (default).



    .. note::

        See :ref:`cuda-memory-management` for more

        details about GPU memory management.

    """
    if device is None:
        device = torch.cuda.current_device()
    device = _get_device_index(device)
    return torch.cuda.cudart().cudaMemGetInfo(device)


def _record_memory_history_legacy(

    enabled: bool,

    record_context=True,

    trace_alloc_max_entries=1,

    trace_alloc_record_context=False,

    device: Union[Device, int] = None,

    record_context_cpp=False,

):
    _C._cuda_record_memory_history_legacy(
        enabled,
        record_context,
        trace_alloc_max_entries,
        trace_alloc_record_context,
        record_context_cpp,
    )


def _record_memory_history(enabled="all", *args, **kwargs):
    """Enable recording of stack traces associated with memory

    allocations, so you can tell what allocated any piece of memory in

    :func:`torch.cuda.memory._snapshot()`.



    In addition too keeping stack traces with each current allocation and free,

    this will also enable recording of a history of all alloc/free events.



    Use :func:`torch.cuda.memory._snapshot()` to retrieve this information,

    and the tools in `_memory_viz.py` to visualize snapshots.



    The Python trace collection is fast (2us per trace), so you may consider

    enabling this on production jobs if you anticipate ever having to debug

    memory issues.



    C++ trace collection is also fast (~50ns/frame), which for many typical programs

    works out to ~2us per trace, but can vary depending on stack depth.



    Args:

        enabled (Literal[None, "state", "all"], optional):

            `None`, disable recording memory history.

            `"state"`, keep information for currenly allocated memory.

            `"all"`, additionally keep a history of all alloc/free calls.

            Defaults to "all".

        context (Literal[None, "state", "alloc", "all"], optional):

            `None`, Do not record any tracebacks.

            `"state"`, Record tracebacks for currently allocated memory.

            `"alloc"`, additionally keep tracebacks for alloc calls.

            `"all"`, additionally keep tracebacks for free calls.

            Defaults to "all".

        stacks (Literal["python", "all"], optional):

            `"python"`, include Python, TorchScript, and inductor frames in tracebacks

            `"all"`, additionally include C++ frames

            Defaults to "all".

        max_entries (int, optional): Keep a maximum of `max_entries`

            alloc/free events in the recorded history recorded.

    """
    if isinstance(enabled, bool):
        return _record_memory_history_legacy(enabled, *args, **kwargs)
    else:
        return _record_memory_history_impl(enabled, *args, **kwargs)


def _record_memory_history_impl(

    enabled: Optional[str] = "all",

    context: Optional[str] = "all",

    stacks: str = "all",

    max_entries: int = sys.maxsize,

    device: Union[Device, int] = None,

):
    _C._cuda_record_memory_history(enabled, context, stacks, max_entries)


_record_memory_history.__signature__ = signature(_record_memory_history_impl)  # type: ignore[attr-defined]


def _snapshot(device: Union[Device, int] = None):
    """Save a snapshot of CUDA memory state at the time it was called.



    The state is represented as a dictionary with the following structure.



    .. code-block:: python



        class Snapshot(TypedDict):

            segments : List[Segment]

            device_traces: List[List[TraceEntry]]



        class Segment(TypedDict):

            # Segments are memory returned from a cudaMalloc call.

            # The size of reserved memory is the sum of all Segments.

            # Segments are cached and reused for future allocations.

            # If the reuse is smaller than the segment, the segment

            # is split into more then one Block.

            # empty_cache() frees Segments that are entirely inactive.

            address: int

            total_size: int #  cudaMalloc'd size of segment

            stream: int

            segment_type: Literal['small', 'large'] # 'large' (>1MB)

            allocated_size: int # size of memory in use

            active_size: int # size of memory in use or in active_awaiting_free state

            blocks : List[Block]



        class Block(TypedDict):

            # A piece of memory returned from the allocator, or

            # current cached but inactive.

            size: int

            requested_size: int # size requested during malloc, may be smaller than

                                # size due to rounding

            address: int

            state: Literal['active_allocated', # used by a tensor

                        'active_awaiting_free', # waiting for another stream to finish using

                                                # this, then it will become free

                        'inactive',] # free for reuse

            frames: List[Frame] # stack trace from where the allocation occurred



        class Frame(TypedDict):

                filename: str

                line: int

                name: str



        class TraceEntry(TypedDict):

            # When `torch.cuda.memory._record_memory_history()` is enabled,

            # the snapshot will contain TraceEntry objects that record each

            # action the allocator took.

            action: Literal[

            'alloc'  # memory allocated

            'free_requested', # the allocated received a call to free memory

            'free_completed', # the memory that was requested to be freed is now

                            # able to be used in future allocation calls

            'segment_alloc', # the caching allocator ask cudaMalloc for more memory

                            # and added it as a segment in its cache

            'segment_free',  # the caching allocator called cudaFree to return memory

                            # to cuda possibly trying free up memory to

                            # allocate more segments or because empty_caches was called

            'oom',          # the allocator threw an OOM exception. 'size' is

                            # the requested number of bytes that did not succeed

            'snapshot'      # the allocator generated a memory snapshot

                            # useful to coorelate a previously taken

                            # snapshot with this trace

            ]

            addr: int # not present for OOM

            frames: List[Frame]

            size: int

            stream: int

            device_free: int # only present for OOM, the amount of

                            # memory cuda still reports to be free



    Returns:

        The Snapshot dictionary object

    """
    return _C._cuda_memorySnapshot()


def _dump_snapshot(filename="dump_snapshot.pickle"):
    """

    Save a pickled version of the `torch.memory._snapshot()` dictionary to a file.



    This file can be opened by the interactive snapshot viewer at pytorch.org/memory_viz



    Args:

        filename (str, optional): Name of the file to create. Defaults to "dump_snapshot.pickle".

    """
    s = _snapshot()
    with open(filename, "wb") as f:
        pickle.dump(s, f)


def _save_segment_usage(filename="output.svg", snapshot=None):
    if snapshot is None:
        snapshot = _snapshot()
    with open(filename, "w") as f:
        f.write(_segments(snapshot))


def _save_memory_usage(filename="output.svg", snapshot=None):
    if snapshot is None:
        snapshot = _snapshot()
    with open(filename, "w") as f:
        f.write(_memory(snapshot))


def _set_allocator_settings(env: str):
    return torch._C._cuda_cudaCachingAllocator_set_allocator_settings(env)


def get_allocator_backend() -> str:
    r"""Return a string describing the active allocator backend as set by

    ``PYTORCH_CUDA_ALLOC_CONF``. Currently available backends are

    ``native`` (PyTorch's native caching allocator) and `cudaMallocAsync``

    (CUDA's built-in asynchronous allocator).



    .. note::

        See :ref:`cuda-memory-management` for details on choosing the allocator backend.

    """
    return torch._C._cuda_getAllocatorBackend()


class _CUDAAllocator:
    r"""Wrapper over internal CUDA memory allocators."""

    def __init__(self, allocator: torch._C._cuda_CUDAAllocator):
        self._allocator = allocator

    def allocator(self):
        return self._allocator


class CUDAPluggableAllocator(_CUDAAllocator):
    r"""CUDA memory allocator loaded from a so file."""

    def __init__(self, path_to_so_file: str, alloc_fn_name: str, free_fn_name: str):
        r"""Memory allocators are compiled in .so files and loaded dynamically using ctypes.



        To change the active allocator use the :func:`torch.memory.cuda.change_current_allocator` function.



        Args:

            path_to_so_file(str): Path in the filesystem to the `.so` file containing

                the allocator functions

            alloc_fn_name(str): Name of the function to perform the memory allocation

                in the so file. The signature must be:

                void* alloc_fn_name(ssize_t size, int device, cudaStream_t stream);

            free_fn_name(str): Name of the function to perform the memory release

                in the so file. The signature must be:

                void free_fn_name(void* ptr, size_t size, cudaStream_t stream);



        .. warning::

            This is currently supported only in unix OSs



        .. note::

            See :ref:`cuda-memory-management` for details on creating and using a custom allocator

        """
        allocator = ctypes.CDLL(path_to_so_file)
        alloc_fn = ctypes.cast(getattr(allocator, alloc_fn_name), ctypes.c_void_p).value
        free_fn = ctypes.cast(getattr(allocator, free_fn_name), ctypes.c_void_p).value
        assert alloc_fn is not None
        assert free_fn is not None
        self._allocator = torch._C._cuda_customAllocator(alloc_fn, free_fn)


def change_current_allocator(allocator: _CUDAAllocator) -> None:
    r"""Change the currently used memory allocator to be the one provided.



    If the current allocator has already been used/initialized, this function will error.





    Args:

        allocator (torch.cuda.memory._CUDAAllocator): allocator to be set as the active one.

    .. note::

        See :ref:`cuda-memory-management` for details on creating and using a custom allocator

    """
    torch._C._cuda_changeCurrentAllocator(allocator.allocator())


def _get_current_allocator() -> _CUDAAllocator:
    r"""Return the allocator being currently used.



    .. note::

        See :ref:`cuda-memory-management` for details on creating and using a custom allocator

    """
    return _CUDAAllocator(torch._C._cuda_getAllocator())