-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathcuda.html
1226 lines (1045 loc) · 83.5 KB
/
cuda.html
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
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta name="robots" content="noindex">
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>torch.cuda — PyTorch 2.5 documentation</title>
<link rel="canonical" href="https://pytorch.org/docs/stable/cuda.html"/>
<link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
<!-- <link rel="stylesheet" href="_static/pygments.css" type="text/css" /> -->
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
<link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
<link rel="stylesheet" href="_static/copybutton.css" type="text/css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/katex.min.css" type="text/css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/katex.min.css" type="text/css" />
<link rel="stylesheet" href="_static/katex-math.css" type="text/css" />
<link rel="stylesheet" href="_static/sphinx-dropdown.css" type="text/css" />
<link rel="stylesheet" href="_static/panels-bootstrap.min.css" type="text/css" />
<link rel="stylesheet" href="_static/css/jit.css" type="text/css" />
<link rel="index" title="Index" href="genindex.html" />
<link rel="search" title="Search" href="search.html" />
<link rel="next" title="StreamContext" href="generated/torch.cuda.StreamContext.html" />
<link rel="prev" title="Stream" href="generated/torch.cpu.Stream.html" />
<!-- Google Tag Manager -->
<script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':
new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],
j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src=
'https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);
})(window,document,'script','dataLayer','GTM-T8XT4PS');</script>
<!-- End Google Tag Manager -->
<script src="_static/js/modernizr.min.js"></script>
<!-- Preload the theme fonts -->
<link rel="preload" href="_static/fonts/FreightSans/freight-sans-book.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/FreightSans/freight-sans-medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/IBMPlexMono/IBMPlexMono-Medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/FreightSans/freight-sans-bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/FreightSans/freight-sans-medium-italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="_static/fonts/IBMPlexMono/IBMPlexMono-SemiBold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<!-- Preload the katex fonts -->
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Math-Italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Main-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Main-Bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Size1-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Size4-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Size2-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Size3-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/fonts/KaTeX_Caligraphic-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.15.2/css/all.css" integrity="sha384-vSIIfh2YWi9wW0r9iZe7RJPrKwp6bG+s9QZMoITbCckVJqGCCRhc+ccxNcdpHuYu" crossorigin="anonymous">
</head>
<div class="container-fluid header-holder tutorials-header" id="header-holder">
<div class="container">
<div class="header-container">
<a class="header-logo" href="https://pytorch.org/" aria-label="PyTorch"></a>
<div class="main-menu">
<ul>
<li class="main-menu-item">
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="with-down-arrow">
Learn
</a>
<div class="resources-dropdown-menu">
<a class="nav-dropdown-item" href="https://pytorch.org/get-started">
<span class=dropdown-title>Get Started</span>
<p>Run PyTorch locally or get started quickly with one of the supported cloud platforms</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/tutorials">
<span class="dropdown-title">Tutorials</span>
<p>Whats new in PyTorch tutorials</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/tutorials/beginner/basics/intro.html">
<span class="dropdown-title">Learn the Basics</span>
<p>Familiarize yourself with PyTorch concepts and modules</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/tutorials/recipes/recipes_index.html">
<span class="dropdown-title">PyTorch Recipes</span>
<p>Bite-size, ready-to-deploy PyTorch code examples</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/tutorials/beginner/introyt.html">
<span class="dropdown-title">Intro to PyTorch - YouTube Series</span>
<p>Master PyTorch basics with our engaging YouTube tutorial series</p>
</a>
</div>
</div>
</li>
<li>
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="with-down-arrow">
Ecosystem
</a>
<div class="resources-dropdown-menu">
<a class="nav-dropdown-item" href="https://pytorch.org/ecosystem">
<span class="dropdown-title">Tools</span>
<p>Learn about the tools and frameworks in the PyTorch Ecosystem</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/#community-module">
<span class=dropdown-title>Community</span>
<p>Join the PyTorch developer community to contribute, learn, and get your questions answered</p>
</a>
<a class="nav-dropdown-item" href="https://discuss.pytorch.org/" target="_blank">
<span class=dropdown-title>Forums</span>
<p>A place to discuss PyTorch code, issues, install, research</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/resources">
<span class=dropdown-title>Developer Resources</span>
<p>Find resources and get questions answered</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/ecosystem/contributor-awards-2023">
<span class="dropdown-title">Contributor Awards - 2023</span>
<p>Award winners announced at this year's PyTorch Conference</p>
</a>
</div>
</div>
</li>
<li>
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="with-down-arrow">
Edge
</a>
<div class="resources-dropdown-menu">
<a class="nav-dropdown-item" href="https://pytorch.org/edge">
<span class="dropdown-title">About PyTorch Edge</span>
<p>Build innovative and privacy-aware AI experiences for edge devices</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/executorch-overview">
<span class="dropdown-title">ExecuTorch</span>
<p>End-to-end solution for enabling on-device inference capabilities across mobile and edge devices</p>
</a>
</div>
</div>
</li>
<li class="main-menu-item">
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="with-down-arrow">
Docs
</a>
<div class="resources-dropdown-menu">
<a class="nav-dropdown-item" href="https://pytorch.org/docs/stable/index.html">
<span class="dropdown-title">PyTorch</span>
<p>Explore the documentation for comprehensive guidance on how to use PyTorch</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/pytorch-domains">
<span class="dropdown-title">PyTorch Domains</span>
<p>Read the PyTorch Domains documentation to learn more about domain-specific libraries</p>
</a>
</div>
</div>
</li>
<li>
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="with-down-arrow">
Blogs & News
</a>
<div class="resources-dropdown-menu">
<a class="nav-dropdown-item" href="https://pytorch.org/blog/">
<span class="dropdown-title">PyTorch Blog</span>
<p>Catch up on the latest technical news and happenings</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/community-blog">
<span class="dropdown-title">Community Blog</span>
<p>Stories from the PyTorch ecosystem</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/videos">
<span class="dropdown-title">Videos</span>
<p>Learn about the latest PyTorch tutorials, new, and more </p>
<a class="nav-dropdown-item" href="https://pytorch.org/community-stories">
<span class="dropdown-title">Community Stories</span>
<p>Learn how our community solves real, everyday machine learning problems with PyTorch</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/events">
<span class="dropdown-title">Events</span>
<p>Find events, webinars, and podcasts</p>
</a>
</div>
</li>
<li>
<div id="resourcesDropdownButton" data-toggle="resources-dropdown" class="resources-dropdown">
<a class="with-down-arrow">
About
</a>
<div class="resources-dropdown-menu">
<a class="nav-dropdown-item" href="https://pytorch.org/foundation">
<span class="dropdown-title">PyTorch Foundation</span>
<p>Learn more about the PyTorch Foundation</p>
</a>
<a class="nav-dropdown-item" href="https://pytorch.org/governing-board">
<span class="dropdown-title">Governing Board</span>
<p></p>
</a>
</div>
</div>
</li>
<li class="main-menu-item">
<div class="no-dropdown">
<a href="https://pytorch.org/join" data-cta="join">
Become a Member
</a>
</div>
</li>
<li>
<div class="main-menu-item">
<a href="https://github.com/pytorch/pytorch" class="github-icon">
</a>
</div>
</li>
<!--- TODO: This block adds the search icon to the nav bar. We will enable it later.
<li>
<div class="main-menu-item">
<a href="https://github.com/pytorch/pytorch" class="search-icon">
</a>
</div>
</li>
--->
</ul>
</div>
<a class="main-menu-open-button" href="#" data-behavior="open-mobile-menu"></a>
</div>
</div>
</div>
<body class="pytorch-body">
<div class="table-of-contents-link-wrapper">
<span>Table of Contents</span>
<a href="#" class="toggle-table-of-contents" data-behavior="toggle-table-of-contents"></a>
</div>
<nav data-toggle="wy-nav-shift" class="pytorch-left-menu" id="pytorch-left-menu">
<div class="pytorch-side-scroll">
<div class="pytorch-menu pytorch-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
<div class="pytorch-left-menu-search">
<div class="version">
<a href='https://pytorch.org/docs/versions.html'>2.5 ▼</a>
</div>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="search.html" method="get">
<input type="text" name="q" placeholder="Search Docs" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div>
<p class="caption" role="heading"><span class="caption-text">Community</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="community/build_ci_governance.html">PyTorch Governance | Build + CI</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/contribution_guide.html">PyTorch Contribution Guide</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/design.html">PyTorch Design Philosophy</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/governance.html">PyTorch Governance | Mechanics</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/persons_of_interest.html">PyTorch Governance | Maintainers</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Developer Notes</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="notes/amp_examples.html">Automatic Mixed Precision examples</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/autograd.html">Autograd mechanics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/broadcasting.html">Broadcasting semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/cpu_threading_torchscript_inference.html">CPU threading and TorchScript inference</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/cuda.html">CUDA semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/custom_operators.html">PyTorch Custom Operators Landing Page</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/ddp.html">Distributed Data Parallel</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/extending.html">Extending PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/extending.func.html">Extending torch.func with autograd.Function</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/faq.html">Frequently Asked Questions</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/fsdp.html">FSDP Notes</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/get_start_xpu.html">Getting Started on Intel GPU</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/gradcheck.html">Gradcheck mechanics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/hip.html">HIP (ROCm) semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/large_scale_deployments.html">Features for large-scale deployments</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/modules.html">Modules</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/mps.html">MPS backend</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/multiprocessing.html">Multiprocessing best practices</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/numerical_accuracy.html">Numerical accuracy</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/randomness.html">Reproducibility</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/serialization.html">Serialization semantics</a></li>
<li class="toctree-l1"><a class="reference internal" href="notes/windows.html">Windows FAQ</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Language Bindings</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="cpp_index.html">C++</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/javadoc/">Javadoc</a></li>
<li class="toctree-l1"><a class="reference internal" href="deploy.html">torch::deploy</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Python API</span></p>
<ul class="current">
<li class="toctree-l1"><a class="reference internal" href="torch.html">torch</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.html">torch.nn</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.functional.html">torch.nn.functional</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensors.html">torch.Tensor</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensor_attributes.html">Tensor Attributes</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensor_view.html">Tensor Views</a></li>
<li class="toctree-l1"><a class="reference internal" href="amp.html">torch.amp</a></li>
<li class="toctree-l1"><a class="reference internal" href="autograd.html">torch.autograd</a></li>
<li class="toctree-l1"><a class="reference internal" href="library.html">torch.library</a></li>
<li class="toctree-l1"><a class="reference internal" href="cpu.html">torch.cpu</a></li>
<li class="toctree-l1 current"><a class="current reference internal" href="#">torch.cuda</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch_cuda_memory.html">Understanding CUDA Memory Usage</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch_cuda_memory.html#generating-a-snapshot">Generating a Snapshot</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch_cuda_memory.html#using-the-visualizer">Using the visualizer</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch_cuda_memory.html#snapshot-api-reference">Snapshot API Reference</a></li>
<li class="toctree-l1"><a class="reference internal" href="mps.html">torch.mps</a></li>
<li class="toctree-l1"><a class="reference internal" href="xpu.html">torch.xpu</a></li>
<li class="toctree-l1"><a class="reference internal" href="mtia.html">torch.mtia</a></li>
<li class="toctree-l1"><a class="reference internal" href="meta.html">Meta device</a></li>
<li class="toctree-l1"><a class="reference internal" href="backends.html">torch.backends</a></li>
<li class="toctree-l1"><a class="reference internal" href="export.html">torch.export</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.html">torch.distributed</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.tensor.html">torch.distributed.tensor</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.algorithms.join.html">torch.distributed.algorithms.join</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.elastic.html">torch.distributed.elastic</a></li>
<li class="toctree-l1"><a class="reference internal" href="fsdp.html">torch.distributed.fsdp</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.tensor.parallel.html">torch.distributed.tensor.parallel</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.optim.html">torch.distributed.optim</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.pipelining.html">torch.distributed.pipelining</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributed.checkpoint.html">torch.distributed.checkpoint</a></li>
<li class="toctree-l1"><a class="reference internal" href="distributions.html">torch.distributions</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch.compiler.html">torch.compiler</a></li>
<li class="toctree-l1"><a class="reference internal" href="fft.html">torch.fft</a></li>
<li class="toctree-l1"><a class="reference internal" href="func.html">torch.func</a></li>
<li class="toctree-l1"><a class="reference internal" href="futures.html">torch.futures</a></li>
<li class="toctree-l1"><a class="reference internal" href="fx.html">torch.fx</a></li>
<li class="toctree-l1"><a class="reference internal" href="fx.experimental.html">torch.fx.experimental</a></li>
<li class="toctree-l1"><a class="reference internal" href="hub.html">torch.hub</a></li>
<li class="toctree-l1"><a class="reference internal" href="jit.html">torch.jit</a></li>
<li class="toctree-l1"><a class="reference internal" href="linalg.html">torch.linalg</a></li>
<li class="toctree-l1"><a class="reference internal" href="monitor.html">torch.monitor</a></li>
<li class="toctree-l1"><a class="reference internal" href="signal.html">torch.signal</a></li>
<li class="toctree-l1"><a class="reference internal" href="special.html">torch.special</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch.overrides.html">torch.overrides</a></li>
<li class="toctree-l1"><a class="reference internal" href="package.html">torch.package</a></li>
<li class="toctree-l1"><a class="reference internal" href="profiler.html">torch.profiler</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.init.html">torch.nn.init</a></li>
<li class="toctree-l1"><a class="reference internal" href="nn.attention.html">torch.nn.attention</a></li>
<li class="toctree-l1"><a class="reference internal" href="onnx.html">torch.onnx</a></li>
<li class="toctree-l1"><a class="reference internal" href="optim.html">torch.optim</a></li>
<li class="toctree-l1"><a class="reference internal" href="complex_numbers.html">Complex Numbers</a></li>
<li class="toctree-l1"><a class="reference internal" href="ddp_comm_hooks.html">DDP Communication Hooks</a></li>
<li class="toctree-l1"><a class="reference internal" href="quantization.html">Quantization</a></li>
<li class="toctree-l1"><a class="reference internal" href="rpc.html">Distributed RPC Framework</a></li>
<li class="toctree-l1"><a class="reference internal" href="random.html">torch.random</a></li>
<li class="toctree-l1"><a class="reference internal" href="masked.html">torch.masked</a></li>
<li class="toctree-l1"><a class="reference internal" href="nested.html">torch.nested</a></li>
<li class="toctree-l1"><a class="reference internal" href="size.html">torch.Size</a></li>
<li class="toctree-l1"><a class="reference internal" href="sparse.html">torch.sparse</a></li>
<li class="toctree-l1"><a class="reference internal" href="storage.html">torch.Storage</a></li>
<li class="toctree-l1"><a class="reference internal" href="testing.html">torch.testing</a></li>
<li class="toctree-l1"><a class="reference internal" href="utils.html">torch.utils</a></li>
<li class="toctree-l1"><a class="reference internal" href="benchmark_utils.html">torch.utils.benchmark</a></li>
<li class="toctree-l1"><a class="reference internal" href="bottleneck.html">torch.utils.bottleneck</a></li>
<li class="toctree-l1"><a class="reference internal" href="checkpoint.html">torch.utils.checkpoint</a></li>
<li class="toctree-l1"><a class="reference internal" href="cpp_extension.html">torch.utils.cpp_extension</a></li>
<li class="toctree-l1"><a class="reference internal" href="data.html">torch.utils.data</a></li>
<li class="toctree-l1"><a class="reference internal" href="deterministic.html">torch.utils.deterministic</a></li>
<li class="toctree-l1"><a class="reference internal" href="jit_utils.html">torch.utils.jit</a></li>
<li class="toctree-l1"><a class="reference internal" href="dlpack.html">torch.utils.dlpack</a></li>
<li class="toctree-l1"><a class="reference internal" href="mobile_optimizer.html">torch.utils.mobile_optimizer</a></li>
<li class="toctree-l1"><a class="reference internal" href="model_zoo.html">torch.utils.model_zoo</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensorboard.html">torch.utils.tensorboard</a></li>
<li class="toctree-l1"><a class="reference internal" href="module_tracker.html">torch.utils.module_tracker</a></li>
<li class="toctree-l1"><a class="reference internal" href="type_info.html">Type Info</a></li>
<li class="toctree-l1"><a class="reference internal" href="named_tensor.html">Named Tensors</a></li>
<li class="toctree-l1"><a class="reference internal" href="name_inference.html">Named Tensors operator coverage</a></li>
<li class="toctree-l1"><a class="reference internal" href="config_mod.html">torch.__config__</a></li>
<li class="toctree-l1"><a class="reference internal" href="future_mod.html">torch.__future__</a></li>
<li class="toctree-l1"><a class="reference internal" href="logging.html">torch._logging</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch_environment_variables.html">Torch Environment Variables</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Libraries</span></p>
<ul>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/audio/stable">torchaudio</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/data">TorchData</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/torchrec">TorchRec</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/serve">TorchServe</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/text/stable">torchtext</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/vision/stable">torchvision</a></li>
<li class="toctree-l1"><a class="reference external" href="https://pytorch.org/xla/">PyTorch on XLA Devices</a></li>
</ul>
</div>
</div>
</nav>
<div class="pytorch-container">
<div class="pytorch-page-level-bar" id="pytorch-page-level-bar">
<div class="pytorch-breadcrumbs-wrapper">
<div role="navigation" aria-label="breadcrumbs navigation">
<ul class="pytorch-breadcrumbs">
<li>
<a href="index.html">
Docs
</a> >
</li>
<li>torch.cuda</li>
<li class="pytorch-breadcrumbs-aside">
<a href="_sources/cuda.rst.txt" rel="nofollow"><img src="_static/images/view-page-source-icon.svg"></a>
</li>
</ul>
</div>
</div>
<div class="pytorch-shortcuts-wrapper" id="pytorch-shortcuts-wrapper">
Shortcuts
</div>
</div>
<section data-toggle="wy-nav-shift" id="pytorch-content-wrap" class="pytorch-content-wrap">
<div class="pytorch-content-left">
<!-- Google Tag Manager (noscript) -->
<noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-T8XT4PS"
height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript>
<!-- End Google Tag Manager (noscript) -->
<div class="rst-content">
<div role="main" class="main-content" itemscope="itemscope" itemtype="http://schema.org/Article">
<article itemprop="articleBody" id="pytorch-article" class="pytorch-article">
<div class="section" id="module-torch.cuda">
<span id="torch-cuda"></span><h1>torch.cuda<a class="headerlink" href="#module-torch.cuda" title="Permalink to this heading">¶</a></h1>
<p>This package adds support for CUDA tensor types.</p>
<p>It implements the same function as CPU tensors, but they utilize
GPUs for computation.</p>
<p>It is lazily initialized, so you can always import it, and use
<a class="reference internal" href="generated/torch.cuda.is_available.html#torch.cuda.is_available" title="torch.cuda.is_available"><code class="xref py py-func docutils literal notranslate"><span class="pre">is_available()</span></code></a> to determine if your system supports CUDA.</p>
<p><a class="reference internal" href="notes/cuda.html#cuda-semantics"><span class="std std-ref">CUDA semantics</span></a> has more details about working with CUDA.</p>
<table class="autosummary longtable docutils colwidths-auto align-default">
<tbody>
<tr class="row-odd"><td><p><p id="torch.cuda.StreamContext"/><a class="reference internal" href="generated/torch.cuda.StreamContext.html#torch.cuda.StreamContext" title="torch.cuda.StreamContext"><code class="xref py py-obj docutils literal notranslate"><span class="pre">StreamContext</span></code></a></p></td>
<td><p>Context-manager that selects a given stream.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.can_device_access_peer"/><a class="reference internal" href="generated/torch.cuda.can_device_access_peer.html#torch.cuda.can_device_access_peer" title="torch.cuda.can_device_access_peer"><code class="xref py py-obj docutils literal notranslate"><span class="pre">can_device_access_peer</span></code></a></p></td>
<td><p>Check if peer access between two devices is possible.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.current_blas_handle"/><a class="reference internal" href="generated/torch.cuda.current_blas_handle.html#torch.cuda.current_blas_handle" title="torch.cuda.current_blas_handle"><code class="xref py py-obj docutils literal notranslate"><span class="pre">current_blas_handle</span></code></a></p></td>
<td><p>Return cublasHandle_t pointer to current cuBLAS handle</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.current_device"/><a class="reference internal" href="generated/torch.cuda.current_device.html#torch.cuda.current_device" title="torch.cuda.current_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">current_device</span></code></a></p></td>
<td><p>Return the index of a currently selected device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.current_stream"/><a class="reference internal" href="generated/torch.cuda.current_stream.html#torch.cuda.current_stream" title="torch.cuda.current_stream"><code class="xref py py-obj docutils literal notranslate"><span class="pre">current_stream</span></code></a></p></td>
<td><p>Return the currently selected <a class="reference internal" href="generated/torch.cuda.Stream.html#torch.cuda.Stream" title="torch.cuda.Stream"><code class="xref py py-class docutils literal notranslate"><span class="pre">Stream</span></code></a> for a given device.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.cudart"/><a class="reference internal" href="generated/torch.cuda.cudart.html#torch.cuda.cudart" title="torch.cuda.cudart"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cudart</span></code></a></p></td>
<td><p>Retrieves the CUDA runtime API module.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.default_stream"/><a class="reference internal" href="generated/torch.cuda.default_stream.html#torch.cuda.default_stream" title="torch.cuda.default_stream"><code class="xref py py-obj docutils literal notranslate"><span class="pre">default_stream</span></code></a></p></td>
<td><p>Return the default <a class="reference internal" href="generated/torch.cuda.Stream.html#torch.cuda.Stream" title="torch.cuda.Stream"><code class="xref py py-class docutils literal notranslate"><span class="pre">Stream</span></code></a> for a given device.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.device"/><a class="reference internal" href="generated/torch.cuda.device.html#torch.cuda.device" title="torch.cuda.device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">device</span></code></a></p></td>
<td><p>Context-manager that changes the selected device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.device_count"/><a class="reference internal" href="generated/torch.cuda.device_count.html#torch.cuda.device_count" title="torch.cuda.device_count"><code class="xref py py-obj docutils literal notranslate"><span class="pre">device_count</span></code></a></p></td>
<td><p>Return the number of GPUs available.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.device_of"/><a class="reference internal" href="generated/torch.cuda.device_of.html#torch.cuda.device_of" title="torch.cuda.device_of"><code class="xref py py-obj docutils literal notranslate"><span class="pre">device_of</span></code></a></p></td>
<td><p>Context-manager that changes the current device to that of given object.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.get_arch_list"/><a class="reference internal" href="generated/torch.cuda.get_arch_list.html#torch.cuda.get_arch_list" title="torch.cuda.get_arch_list"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_arch_list</span></code></a></p></td>
<td><p>Return list CUDA architectures this library was compiled for.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.get_device_capability"/><a class="reference internal" href="generated/torch.cuda.get_device_capability.html#torch.cuda.get_device_capability" title="torch.cuda.get_device_capability"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_device_capability</span></code></a></p></td>
<td><p>Get the cuda capability of a device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.get_device_name"/><a class="reference internal" href="generated/torch.cuda.get_device_name.html#torch.cuda.get_device_name" title="torch.cuda.get_device_name"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_device_name</span></code></a></p></td>
<td><p>Get the name of a device.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.get_device_properties"/><a class="reference internal" href="generated/torch.cuda.get_device_properties.html#torch.cuda.get_device_properties" title="torch.cuda.get_device_properties"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_device_properties</span></code></a></p></td>
<td><p>Get the properties of a device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.get_gencode_flags"/><a class="reference internal" href="generated/torch.cuda.get_gencode_flags.html#torch.cuda.get_gencode_flags" title="torch.cuda.get_gencode_flags"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_gencode_flags</span></code></a></p></td>
<td><p>Return NVCC gencode flags this library was compiled with.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.get_sync_debug_mode"/><a class="reference internal" href="generated/torch.cuda.get_sync_debug_mode.html#torch.cuda.get_sync_debug_mode" title="torch.cuda.get_sync_debug_mode"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_sync_debug_mode</span></code></a></p></td>
<td><p>Return current value of debug mode for cuda synchronizing operations.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.init"/><a class="reference internal" href="generated/torch.cuda.init.html#torch.cuda.init" title="torch.cuda.init"><code class="xref py py-obj docutils literal notranslate"><span class="pre">init</span></code></a></p></td>
<td><p>Initialize PyTorch's CUDA state.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.ipc_collect"/><a class="reference internal" href="generated/torch.cuda.ipc_collect.html#torch.cuda.ipc_collect" title="torch.cuda.ipc_collect"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ipc_collect</span></code></a></p></td>
<td><p>Force collects GPU memory after it has been released by CUDA IPC.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.is_available"/><a class="reference internal" href="generated/torch.cuda.is_available.html#torch.cuda.is_available" title="torch.cuda.is_available"><code class="xref py py-obj docutils literal notranslate"><span class="pre">is_available</span></code></a></p></td>
<td><p>Return a bool indicating if CUDA is currently available.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.is_initialized"/><a class="reference internal" href="generated/torch.cuda.is_initialized.html#torch.cuda.is_initialized" title="torch.cuda.is_initialized"><code class="xref py py-obj docutils literal notranslate"><span class="pre">is_initialized</span></code></a></p></td>
<td><p>Return whether PyTorch's CUDA state has been initialized.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.memory_usage"/><a class="reference internal" href="generated/torch.cuda.memory_usage.html#torch.cuda.memory_usage" title="torch.cuda.memory_usage"><code class="xref py py-obj docutils literal notranslate"><span class="pre">memory_usage</span></code></a></p></td>
<td><p>Return the percent of time over the past sample period during which global (device) memory was being read or written as given by <cite>nvidia-smi</cite>.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.set_device"/><a class="reference internal" href="generated/torch.cuda.set_device.html#torch.cuda.set_device" title="torch.cuda.set_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">set_device</span></code></a></p></td>
<td><p>Set the current device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.set_stream"/><a class="reference internal" href="generated/torch.cuda.set_stream.html#torch.cuda.set_stream" title="torch.cuda.set_stream"><code class="xref py py-obj docutils literal notranslate"><span class="pre">set_stream</span></code></a></p></td>
<td><p>Set the current stream.This is a wrapper API to set the stream.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.set_sync_debug_mode"/><a class="reference internal" href="generated/torch.cuda.set_sync_debug_mode.html#torch.cuda.set_sync_debug_mode" title="torch.cuda.set_sync_debug_mode"><code class="xref py py-obj docutils literal notranslate"><span class="pre">set_sync_debug_mode</span></code></a></p></td>
<td><p>Set the debug mode for cuda synchronizing operations.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.stream"/><a class="reference internal" href="generated/torch.cuda.stream.html#torch.cuda.stream" title="torch.cuda.stream"><code class="xref py py-obj docutils literal notranslate"><span class="pre">stream</span></code></a></p></td>
<td><p>Wrap around the Context-manager StreamContext that selects a given stream.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.synchronize"/><a class="reference internal" href="generated/torch.cuda.synchronize.html#torch.cuda.synchronize" title="torch.cuda.synchronize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">synchronize</span></code></a></p></td>
<td><p>Wait for all kernels in all streams on a CUDA device to complete.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.utilization"/><a class="reference internal" href="generated/torch.cuda.utilization.html#torch.cuda.utilization" title="torch.cuda.utilization"><code class="xref py py-obj docutils literal notranslate"><span class="pre">utilization</span></code></a></p></td>
<td><p>Return the percent of time over the past sample period during which one or more kernels was executing on the GPU as given by <cite>nvidia-smi</cite>.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.temperature"/><a class="reference internal" href="generated/torch.cuda.temperature.html#torch.cuda.temperature" title="torch.cuda.temperature"><code class="xref py py-obj docutils literal notranslate"><span class="pre">temperature</span></code></a></p></td>
<td><p>Return the average temperature of the GPU sensor in Degrees C (Centigrades).</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.power_draw"/><a class="reference internal" href="generated/torch.cuda.power_draw.html#torch.cuda.power_draw" title="torch.cuda.power_draw"><code class="xref py py-obj docutils literal notranslate"><span class="pre">power_draw</span></code></a></p></td>
<td><p>Return the average power draw of the GPU sensor in mW (MilliWatts)</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.clock_rate"/><a class="reference internal" href="generated/torch.cuda.clock_rate.html#torch.cuda.clock_rate" title="torch.cuda.clock_rate"><code class="xref py py-obj docutils literal notranslate"><span class="pre">clock_rate</span></code></a></p></td>
<td><p>Return the clock speed of the GPU SM in Hz Hertz over the past sample period as given by <cite>nvidia-smi</cite>.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.OutOfMemoryError"/><a class="reference internal" href="generated/torch.cuda.OutOfMemoryError.html#torch.cuda.OutOfMemoryError" title="torch.cuda.OutOfMemoryError"><code class="xref py py-obj docutils literal notranslate"><span class="pre">OutOfMemoryError</span></code></a></p></td>
<td><p>Exception raised when device is out of memory</p></td>
</tr>
</tbody>
</table>
<div class="section" id="random-number-generator">
<h2>Random Number Generator<a class="headerlink" href="#random-number-generator" title="Permalink to this heading">¶</a></h2>
<table class="autosummary longtable docutils colwidths-auto align-default">
<tbody>
<tr class="row-odd"><td><p><p id="torch.cuda.get_rng_state"/><a class="reference internal" href="generated/torch.cuda.get_rng_state.html#torch.cuda.get_rng_state" title="torch.cuda.get_rng_state"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_rng_state</span></code></a></p></td>
<td><p>Return the random number generator state of the specified GPU as a ByteTensor.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.get_rng_state_all"/><a class="reference internal" href="generated/torch.cuda.get_rng_state_all.html#torch.cuda.get_rng_state_all" title="torch.cuda.get_rng_state_all"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_rng_state_all</span></code></a></p></td>
<td><p>Return a list of ByteTensor representing the random number states of all devices.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.set_rng_state"/><a class="reference internal" href="generated/torch.cuda.set_rng_state.html#torch.cuda.set_rng_state" title="torch.cuda.set_rng_state"><code class="xref py py-obj docutils literal notranslate"><span class="pre">set_rng_state</span></code></a></p></td>
<td><p>Set the random number generator state of the specified GPU.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.set_rng_state_all"/><a class="reference internal" href="generated/torch.cuda.set_rng_state_all.html#torch.cuda.set_rng_state_all" title="torch.cuda.set_rng_state_all"><code class="xref py py-obj docutils literal notranslate"><span class="pre">set_rng_state_all</span></code></a></p></td>
<td><p>Set the random number generator state of all devices.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.manual_seed"/><a class="reference internal" href="generated/torch.cuda.manual_seed.html#torch.cuda.manual_seed" title="torch.cuda.manual_seed"><code class="xref py py-obj docutils literal notranslate"><span class="pre">manual_seed</span></code></a></p></td>
<td><p>Set the seed for generating random numbers for the current GPU.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.manual_seed_all"/><a class="reference internal" href="generated/torch.cuda.manual_seed_all.html#torch.cuda.manual_seed_all" title="torch.cuda.manual_seed_all"><code class="xref py py-obj docutils literal notranslate"><span class="pre">manual_seed_all</span></code></a></p></td>
<td><p>Set the seed for generating random numbers on all GPUs.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.seed"/><a class="reference internal" href="generated/torch.cuda.seed.html#torch.cuda.seed" title="torch.cuda.seed"><code class="xref py py-obj docutils literal notranslate"><span class="pre">seed</span></code></a></p></td>
<td><p>Set the seed for generating random numbers to a random number for the current GPU.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.seed_all"/><a class="reference internal" href="generated/torch.cuda.seed_all.html#torch.cuda.seed_all" title="torch.cuda.seed_all"><code class="xref py py-obj docutils literal notranslate"><span class="pre">seed_all</span></code></a></p></td>
<td><p>Set the seed for generating random numbers to a random number on all GPUs.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.initial_seed"/><a class="reference internal" href="generated/torch.cuda.initial_seed.html#torch.cuda.initial_seed" title="torch.cuda.initial_seed"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initial_seed</span></code></a></p></td>
<td><p>Return the current random seed of the current GPU.</p></td>
</tr>
</tbody>
</table>
</div>
<div class="section" id="communication-collectives">
<h2>Communication collectives<a class="headerlink" href="#communication-collectives" title="Permalink to this heading">¶</a></h2>
<table class="autosummary longtable docutils colwidths-auto align-default">
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="generated/torch.cuda.comm.broadcast.html#torch.cuda.comm.broadcast" title="torch.cuda.comm.broadcast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">comm.broadcast</span></code></a></p></td>
<td><p>Broadcasts a tensor to specified GPU devices.</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="generated/torch.cuda.comm.broadcast_coalesced.html#torch.cuda.comm.broadcast_coalesced" title="torch.cuda.comm.broadcast_coalesced"><code class="xref py py-obj docutils literal notranslate"><span class="pre">comm.broadcast_coalesced</span></code></a></p></td>
<td><p>Broadcast a sequence of tensors to the specified GPUs.</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="generated/torch.cuda.comm.reduce_add.html#torch.cuda.comm.reduce_add" title="torch.cuda.comm.reduce_add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">comm.reduce_add</span></code></a></p></td>
<td><p>Sum tensors from multiple GPUs.</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="generated/torch.cuda.comm.scatter.html#torch.cuda.comm.scatter" title="torch.cuda.comm.scatter"><code class="xref py py-obj docutils literal notranslate"><span class="pre">comm.scatter</span></code></a></p></td>
<td><p>Scatters tensor across multiple GPUs.</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="generated/torch.cuda.comm.gather.html#torch.cuda.comm.gather" title="torch.cuda.comm.gather"><code class="xref py py-obj docutils literal notranslate"><span class="pre">comm.gather</span></code></a></p></td>
<td><p>Gathers tensors from multiple GPU devices.</p></td>
</tr>
</tbody>
</table>
</div>
<div class="section" id="streams-and-events">
<h2>Streams and events<a class="headerlink" href="#streams-and-events" title="Permalink to this heading">¶</a></h2>
<table class="autosummary longtable docutils colwidths-auto align-default">
<tbody>
<tr class="row-odd"><td><p><p id="torch.cuda.Stream"/><a class="reference internal" href="generated/torch.cuda.Stream.html#torch.cuda.Stream" title="torch.cuda.Stream"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Stream</span></code></a></p></td>
<td><p>Wrapper around a CUDA stream.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.ExternalStream"/><a class="reference internal" href="generated/torch.cuda.ExternalStream.html#torch.cuda.ExternalStream" title="torch.cuda.ExternalStream"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ExternalStream</span></code></a></p></td>
<td><p>Wrapper around an externally allocated CUDA stream.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.Event"/><a class="reference internal" href="generated/torch.cuda.Event.html#torch.cuda.Event" title="torch.cuda.Event"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Event</span></code></a></p></td>
<td><p>Wrapper around a CUDA event.</p></td>
</tr>
</tbody>
</table>
</div>
<div class="section" id="graphs-beta">
<h2>Graphs (beta)<a class="headerlink" href="#graphs-beta" title="Permalink to this heading">¶</a></h2>
<table class="autosummary longtable docutils colwidths-auto align-default">
<tbody>
<tr class="row-odd"><td><p><p id="torch.cuda.is_current_stream_capturing"/><a class="reference internal" href="generated/torch.cuda.is_current_stream_capturing.html#torch.cuda.is_current_stream_capturing" title="torch.cuda.is_current_stream_capturing"><code class="xref py py-obj docutils literal notranslate"><span class="pre">is_current_stream_capturing</span></code></a></p></td>
<td><p>Return True if CUDA graph capture is underway on the current CUDA stream, False otherwise.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.graph_pool_handle"/><a class="reference internal" href="generated/torch.cuda.graph_pool_handle.html#torch.cuda.graph_pool_handle" title="torch.cuda.graph_pool_handle"><code class="xref py py-obj docutils literal notranslate"><span class="pre">graph_pool_handle</span></code></a></p></td>
<td><p>Return an opaque token representing the id of a graph memory pool.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.CUDAGraph"/><a class="reference internal" href="generated/torch.cuda.CUDAGraph.html#torch.cuda.CUDAGraph" title="torch.cuda.CUDAGraph"><code class="xref py py-obj docutils literal notranslate"><span class="pre">CUDAGraph</span></code></a></p></td>
<td><p>Wrapper around a CUDA graph.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.graph"/><a class="reference internal" href="generated/torch.cuda.graph.html#torch.cuda.graph" title="torch.cuda.graph"><code class="xref py py-obj docutils literal notranslate"><span class="pre">graph</span></code></a></p></td>
<td><p>Context-manager that captures CUDA work into a <a class="reference internal" href="generated/torch.cuda.CUDAGraph.html#torch.cuda.CUDAGraph" title="torch.cuda.CUDAGraph"><code class="xref py py-class docutils literal notranslate"><span class="pre">torch.cuda.CUDAGraph</span></code></a> object for later replay.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.make_graphed_callables"/><a class="reference internal" href="generated/torch.cuda.make_graphed_callables.html#torch.cuda.make_graphed_callables" title="torch.cuda.make_graphed_callables"><code class="xref py py-obj docutils literal notranslate"><span class="pre">make_graphed_callables</span></code></a></p></td>
<td><p>Accept callables (functions or <a class="reference internal" href="generated/torch.nn.Module.html#torch.nn.Module" title="torch.nn.Module"><code class="xref py py-class docutils literal notranslate"><span class="pre">nn.Module</span></code></a>s) and returns graphed versions.</p></td>
</tr>
</tbody>
</table>
</div>
<div class="section" id="memory-management">
<span id="cuda-memory-management-api"></span><h2>Memory management<a class="headerlink" href="#memory-management" title="Permalink to this heading">¶</a></h2>
<table class="autosummary longtable docutils colwidths-auto align-default">
<tbody>
<tr class="row-odd"><td><p><p id="torch.cuda.empty_cache"/><a class="reference internal" href="generated/torch.cuda.empty_cache.html#torch.cuda.empty_cache" title="torch.cuda.empty_cache"><code class="xref py py-obj docutils literal notranslate"><span class="pre">empty_cache</span></code></a></p></td>
<td><p>Release all unoccupied cached memory currently held by the caching allocator so that those can be used in other GPU application and visible in <cite>nvidia-smi</cite>.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.list_gpu_processes"/><a class="reference internal" href="generated/torch.cuda.list_gpu_processes.html#torch.cuda.list_gpu_processes" title="torch.cuda.list_gpu_processes"><code class="xref py py-obj docutils literal notranslate"><span class="pre">list_gpu_processes</span></code></a></p></td>
<td><p>Return a human-readable printout of the running processes and their GPU memory use for a given device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.mem_get_info"/><a class="reference internal" href="generated/torch.cuda.mem_get_info.html#torch.cuda.mem_get_info" title="torch.cuda.mem_get_info"><code class="xref py py-obj docutils literal notranslate"><span class="pre">mem_get_info</span></code></a></p></td>
<td><p>Return the global free and total GPU memory for a given device using cudaMemGetInfo.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.memory_stats"/><a class="reference internal" href="generated/torch.cuda.memory_stats.html#torch.cuda.memory_stats" title="torch.cuda.memory_stats"><code class="xref py py-obj docutils literal notranslate"><span class="pre">memory_stats</span></code></a></p></td>
<td><p>Return a dictionary of CUDA memory allocator statistics for a given device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.memory_summary"/><a class="reference internal" href="generated/torch.cuda.memory_summary.html#torch.cuda.memory_summary" title="torch.cuda.memory_summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">memory_summary</span></code></a></p></td>
<td><p>Return a human-readable printout of the current memory allocator statistics for a given device.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.memory_snapshot"/><a class="reference internal" href="generated/torch.cuda.memory_snapshot.html#torch.cuda.memory_snapshot" title="torch.cuda.memory_snapshot"><code class="xref py py-obj docutils literal notranslate"><span class="pre">memory_snapshot</span></code></a></p></td>
<td><p>Return a snapshot of the CUDA memory allocator state across all devices.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.memory_allocated"/><a class="reference internal" href="generated/torch.cuda.memory_allocated.html#torch.cuda.memory_allocated" title="torch.cuda.memory_allocated"><code class="xref py py-obj docutils literal notranslate"><span class="pre">memory_allocated</span></code></a></p></td>
<td><p>Return the current GPU memory occupied by tensors in bytes for a given device.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.max_memory_allocated"/><a class="reference internal" href="generated/torch.cuda.max_memory_allocated.html#torch.cuda.max_memory_allocated" title="torch.cuda.max_memory_allocated"><code class="xref py py-obj docutils literal notranslate"><span class="pre">max_memory_allocated</span></code></a></p></td>
<td><p>Return the maximum GPU memory occupied by tensors in bytes for a given device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.reset_max_memory_allocated"/><a class="reference internal" href="generated/torch.cuda.reset_max_memory_allocated.html#torch.cuda.reset_max_memory_allocated" title="torch.cuda.reset_max_memory_allocated"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_max_memory_allocated</span></code></a></p></td>
<td><p>Reset the starting point in tracking maximum GPU memory occupied by tensors for a given device.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.memory_reserved"/><a class="reference internal" href="generated/torch.cuda.memory_reserved.html#torch.cuda.memory_reserved" title="torch.cuda.memory_reserved"><code class="xref py py-obj docutils literal notranslate"><span class="pre">memory_reserved</span></code></a></p></td>
<td><p>Return the current GPU memory managed by the caching allocator in bytes for a given device.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.max_memory_reserved"/><a class="reference internal" href="generated/torch.cuda.max_memory_reserved.html#torch.cuda.max_memory_reserved" title="torch.cuda.max_memory_reserved"><code class="xref py py-obj docutils literal notranslate"><span class="pre">max_memory_reserved</span></code></a></p></td>
<td><p>Return the maximum GPU memory managed by the caching allocator in bytes for a given device.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.set_per_process_memory_fraction"/><a class="reference internal" href="generated/torch.cuda.set_per_process_memory_fraction.html#torch.cuda.set_per_process_memory_fraction" title="torch.cuda.set_per_process_memory_fraction"><code class="xref py py-obj docutils literal notranslate"><span class="pre">set_per_process_memory_fraction</span></code></a></p></td>
<td><p>Set memory fraction for a process.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.memory_cached"/><a class="reference internal" href="generated/torch.cuda.memory_cached.html#torch.cuda.memory_cached" title="torch.cuda.memory_cached"><code class="xref py py-obj docutils literal notranslate"><span class="pre">memory_cached</span></code></a></p></td>
<td><p>Deprecated; see <a class="reference internal" href="generated/torch.cuda.memory_reserved.html#torch.cuda.memory_reserved" title="torch.cuda.memory_reserved"><code class="xref py py-func docutils literal notranslate"><span class="pre">memory_reserved()</span></code></a>.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.max_memory_cached"/><a class="reference internal" href="generated/torch.cuda.max_memory_cached.html#torch.cuda.max_memory_cached" title="torch.cuda.max_memory_cached"><code class="xref py py-obj docutils literal notranslate"><span class="pre">max_memory_cached</span></code></a></p></td>
<td><p>Deprecated; see <a class="reference internal" href="generated/torch.cuda.max_memory_reserved.html#torch.cuda.max_memory_reserved" title="torch.cuda.max_memory_reserved"><code class="xref py py-func docutils literal notranslate"><span class="pre">max_memory_reserved()</span></code></a>.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.reset_max_memory_cached"/><a class="reference internal" href="generated/torch.cuda.reset_max_memory_cached.html#torch.cuda.reset_max_memory_cached" title="torch.cuda.reset_max_memory_cached"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_max_memory_cached</span></code></a></p></td>
<td><p>Reset the starting point in tracking maximum GPU memory managed by the caching allocator for a given device.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.reset_peak_memory_stats"/><a class="reference internal" href="generated/torch.cuda.reset_peak_memory_stats.html#torch.cuda.reset_peak_memory_stats" title="torch.cuda.reset_peak_memory_stats"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_peak_memory_stats</span></code></a></p></td>
<td><p>Reset the "peak" stats tracked by the CUDA memory allocator.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.caching_allocator_alloc"/><a class="reference internal" href="generated/torch.cuda.caching_allocator_alloc.html#torch.cuda.caching_allocator_alloc" title="torch.cuda.caching_allocator_alloc"><code class="xref py py-obj docutils literal notranslate"><span class="pre">caching_allocator_alloc</span></code></a></p></td>
<td><p>Perform a memory allocation using the CUDA memory allocator.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.caching_allocator_delete"/><a class="reference internal" href="generated/torch.cuda.caching_allocator_delete.html#torch.cuda.caching_allocator_delete" title="torch.cuda.caching_allocator_delete"><code class="xref py py-obj docutils literal notranslate"><span class="pre">caching_allocator_delete</span></code></a></p></td>
<td><p>Delete memory allocated using the CUDA memory allocator.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.get_allocator_backend"/><a class="reference internal" href="generated/torch.cuda.get_allocator_backend.html#torch.cuda.get_allocator_backend" title="torch.cuda.get_allocator_backend"><code class="xref py py-obj docutils literal notranslate"><span class="pre">get_allocator_backend</span></code></a></p></td>
<td><p>Return a string describing the active allocator backend as set by <code class="docutils literal notranslate"><span class="pre">PYTORCH_CUDA_ALLOC_CONF</span></code>.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.CUDAPluggableAllocator"/><a class="reference internal" href="generated/torch.cuda.CUDAPluggableAllocator.html#torch.cuda.CUDAPluggableAllocator" title="torch.cuda.CUDAPluggableAllocator"><code class="xref py py-obj docutils literal notranslate"><span class="pre">CUDAPluggableAllocator</span></code></a></p></td>
<td><p>CUDA memory allocator loaded from a so file.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.change_current_allocator"/><a class="reference internal" href="generated/torch.cuda.change_current_allocator.html#torch.cuda.change_current_allocator" title="torch.cuda.change_current_allocator"><code class="xref py py-obj docutils literal notranslate"><span class="pre">change_current_allocator</span></code></a></p></td>
<td><p>Change the currently used memory allocator to be the one provided.</p></td>
</tr>
<tr class="row-even"><td><p><p id="torch.cuda.MemPool"/><a class="reference internal" href="generated/torch.cuda.MemPool.html#torch.cuda.MemPool" title="torch.cuda.MemPool"><code class="xref py py-obj docutils literal notranslate"><span class="pre">MemPool</span></code></a></p></td>
<td><p>MemPool represents a pool of memory in a caching allocator.</p></td>
</tr>
<tr class="row-odd"><td><p><p id="torch.cuda.MemPoolContext"/><a class="reference internal" href="generated/torch.cuda.MemPoolContext.html#torch.cuda.MemPoolContext" title="torch.cuda.MemPoolContext"><code class="xref py py-obj docutils literal notranslate"><span class="pre">MemPoolContext</span></code></a></p></td>
<td><p>MemPoolContext holds the currently active pool and stashes the previous pool.</p></td>
</tr>
</tbody>
</table>
<dl class="py class">
<dt class="sig sig-object py" id="torch.cuda.use_mem_pool">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">torch.cuda.</span></span><span class="sig-name descname"><span class="pre">use_mem_pool</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">pool</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">device</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/cuda/memory.html#use_mem_pool"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#torch.cuda.use_mem_pool" title="Permalink to this definition">¶</a></dt>
<dd><p>A context manager that routes allocations to a given pool.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>pool</strong> (<a class="reference internal" href="generated/torch.cuda.MemPool.html#torch.cuda.MemPool" title="torch.cuda.MemPool"><em>torch.cuda.MemPool</em></a>) – a MemPool object to be made active so that
allocations route to this pool.</p></li>
<li><p><strong>device</strong> (<a class="reference internal" href="tensor_attributes.html#torch.device" title="torch.device"><em>torch.device</em></a><em> or </em><a class="reference external" href="https://docs.python.org/3/library/functions.html#int" title="(in Python v3.13)"><em>int</em></a><em>, </em><em>optional</em>) – selected device. Uses MemPool on
the current device, given by <a class="reference internal" href="generated/torch.cuda.current_device.html#torch.cuda.current_device" title="torch.cuda.current_device"><code class="xref py py-func docutils literal notranslate"><span class="pre">current_device()</span></code></a>,
if <a class="reference internal" href="generated/torch.cuda.device.html#torch.cuda.device" title="torch.cuda.device"><code class="xref py py-attr docutils literal notranslate"><span class="pre">device</span></code></a> is <code class="docutils literal notranslate"><span class="pre">None</span></code> (default).</p></li>
</ul>
</dd>
</dl>
</dd></dl>
</div>
<div class="section" id="nvidia-tools-extension-nvtx">
<h2>NVIDIA Tools Extension (NVTX)<a class="headerlink" href="#nvidia-tools-extension-nvtx" title="Permalink to this heading">¶</a></h2>
<table class="autosummary longtable docutils colwidths-auto align-default">
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="generated/torch.cuda.nvtx.mark.html#torch.cuda.nvtx.mark" title="torch.cuda.nvtx.mark"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nvtx.mark</span></code></a></p></td>
<td><p>Describe an instantaneous event that occurred at some point.</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="generated/torch.cuda.nvtx.range_push.html#torch.cuda.nvtx.range_push" title="torch.cuda.nvtx.range_push"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nvtx.range_push</span></code></a></p></td>
<td><p>Push a range onto a stack of nested range span.</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="generated/torch.cuda.nvtx.range_pop.html#torch.cuda.nvtx.range_pop" title="torch.cuda.nvtx.range_pop"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nvtx.range_pop</span></code></a></p></td>
<td><p>Pop a range off of a stack of nested range spans.</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="generated/torch.cuda.nvtx.range.html#torch.cuda.nvtx.range" title="torch.cuda.nvtx.range"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nvtx.range</span></code></a></p></td>
<td><p>Context manager / decorator that pushes an NVTX range at the beginning of its scope, and pops it at the end.</p></td>
</tr>
</tbody>
</table>
</div>
<div class="section" id="jiterator-beta">
<h2>Jiterator (beta)<a class="headerlink" href="#jiterator-beta" title="Permalink to this heading">¶</a></h2>
<table class="autosummary longtable docutils colwidths-auto align-default">
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="generated/torch.cuda.jiterator._create_jit_fn.html#torch.cuda.jiterator._create_jit_fn" title="torch.cuda.jiterator._create_jit_fn"><code class="xref py py-obj docutils literal notranslate"><span class="pre">jiterator._create_jit_fn</span></code></a></p></td>
<td><p>Create a jiterator-generated cuda kernel for an elementwise op.</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="generated/torch.cuda.jiterator._create_multi_output_jit_fn.html#torch.cuda.jiterator._create_multi_output_jit_fn" title="torch.cuda.jiterator._create_multi_output_jit_fn"><code class="xref py py-obj docutils literal notranslate"><span class="pre">jiterator._create_multi_output_jit_fn</span></code></a></p></td>
<td><p>Create a jiterator-generated cuda kernel for an elementwise op that supports returning one or more outputs.</p></td>
</tr>
</tbody>
</table>
</div>
<div class="section" id="tunableop">
<h2>TunableOp<a class="headerlink" href="#tunableop" title="Permalink to this heading">¶</a></h2>
<p>Some operations could be implemented using more than one library or more than
one technique. For example, a GEMM could be implemented for CUDA or ROCm using
either the cublas/cublasLt libraries or hipblas/hipblasLt libraries,
respectively. How does one know which implementation is the fastest and should
be chosen? That’s what TunableOp provides. Certain operators have been
implemented using multiple strategies as Tunable Operators. At runtime, all
strategies are profiled and the fastest is selected for all subsequent
operations.</p>
<p>See the <a class="reference internal" href="cuda.tunable.html"><span class="doc">documentation</span></a> for information on how to use it.</p>
<div class="toctree-wrapper compound">
</div>
</div>
<div class="section" id="stream-sanitizer-prototype">
<h2>Stream Sanitizer (prototype)<a class="headerlink" href="#stream-sanitizer-prototype" title="Permalink to this heading">¶</a></h2>
<p>CUDA Sanitizer is a prototype tool for detecting synchronization errors between streams in PyTorch.
See the <a class="reference internal" href="cuda._sanitizer.html"><span class="doc">documentation</span></a> for information on how to use it.</p>
<div class="toctree-wrapper compound">
</div>
<span class="target" id="module-torch.cuda.comm"></span><span class="target" id="module-torch.cuda.error"></span><span class="target" id="module-torch.cuda.gds"></span><span class="target" id="module-torch.cuda.graphs"></span><span class="target" id="module-torch.cuda.jiterator"></span><span class="target" id="module-torch.cuda.memory"></span><span class="target" id="module-torch.cuda.nccl"></span><span class="target" id="module-torch.cuda.nvtx"></span><span class="target" id="module-torch.cuda.profiler"></span><span class="target" id="module-torch.cuda.random"></span><span class="target" id="module-torch.cuda.sparse"></span><span class="target" id="module-torch.cuda.streams"></span></div>
</div>
</article>
</div>
<footer>
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
<a href="generated/torch.cuda.StreamContext.html" class="btn btn-neutral float-right" title="StreamContext" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
<a href="generated/torch.cpu.Stream.html" class="btn btn-neutral" title="Stream" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
</div>
<hr>
<div role="contentinfo">
<p>
© Copyright 2023, PyTorch Contributors.
</p>
</div>
<div>
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</div>
</footer>
</div>
<script>
var match = window.location.href.match(/\/_[a-zA-Z0-9_]*.html|_dynamo/gi);
var url = window.location.href.lastIndexOf(match[match.length-1]);
if (url)
{
var div = '<div class="admonition note"><p class="admonition-title">Note</p><p><i class="fa fa-exclamation-circle" aria-hidden="true"> </i> This page describes an internal API which is not intended to be used outside of the PyTorch codebase and can be modified or removed without notice.</p></div>'
document.getElementById("pytorch-article").insertAdjacentHTML('afterBegin', div)
}
</script>
</div>
<div class="pytorch-content-right" id="pytorch-content-right">
<div class="pytorch-right-menu" id="pytorch-right-menu">
<div class="pytorch-side-scroll" id="pytorch-side-scroll-right">
<ul>
<li><a class="reference internal" href="#">torch.cuda</a><ul>
<li><a class="reference internal" href="#random-number-generator">Random Number Generator</a></li>
<li><a class="reference internal" href="#communication-collectives">Communication collectives</a></li>
<li><a class="reference internal" href="#streams-and-events">Streams and events</a></li>
<li><a class="reference internal" href="#graphs-beta">Graphs (beta)</a></li>
<li><a class="reference internal" href="#memory-management">Memory management</a><ul>
<li><a class="reference internal" href="#torch.cuda.use_mem_pool"><code class="docutils literal notranslate"><span class="pre">use_mem_pool</span></code></a></li>
</ul>
</li>
<li><a class="reference internal" href="#nvidia-tools-extension-nvtx">NVIDIA Tools Extension (NVTX)</a></li>
<li><a class="reference internal" href="#jiterator-beta">Jiterator (beta)</a></li>
<li><a class="reference internal" href="#tunableop">TunableOp</a></li>
<li><a class="reference internal" href="#stream-sanitizer-prototype">Stream Sanitizer (prototype)</a></li>
</ul>
</li>
</ul>
</div>
</div>
</div>
</section>
</div>
<script type="text/javascript" id="documentation_options" data-url_root="./" src="_static/documentation_options.js"></script>
<script data-url_root="./" id="documentation_options" src="_static/documentation_options.js"></script>
<script src="_static/jquery.js"></script>
<script src="_static/underscore.js"></script>
<script src="_static/_sphinx_javascript_frameworks_compat.js"></script>
<script src="_static/doctools.js"></script>
<script src="_static/sphinx_highlight.js"></script>
<script src="_static/clipboard.min.js"></script>
<script src="_static/copybutton.js"></script>
<script type="text/javascript" src="_static/js/vendor/popper.min.js"></script>
<script type="text/javascript" src="_static/js/vendor/bootstrap.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/list.js/1.5.0/list.min.js"></script>
<script type="text/javascript" src="_static/js/theme.js"></script>
<script type="text/javascript">
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
<script script type="text/javascript">
var collapsedSections = ['Developer Notes', 'Language Bindings', 'Libraries', 'Community'];
</script>
<img height="1" width="1" style="border-style:none;" alt="" src="https://www.googleadservices.com/pagead/conversion/795629140/?label=txkmCPmdtosBENSssfsC&guid=ON&script=0"/>
<!-- Begin Footer -->
<div class="container-fluid docs-tutorials-resources" id="docs-tutorials-resources">
<div class="container">
<div class="row">
<div class="col-md-4 text-center">
<h2>Docs</h2>
<p>Access comprehensive developer documentation for PyTorch</p>
<a class="with-right-arrow" href="https://pytorch.org/docs/stable/index.html">View Docs</a>
</div>
<div class="col-md-4 text-center">
<h2>Tutorials</h2>
<p>Get in-depth tutorials for beginners and advanced developers</p>
<a class="with-right-arrow" href="https://pytorch.org/tutorials">View Tutorials</a>
</div>
<div class="col-md-4 text-center">
<h2>Resources</h2>
<p>Find development resources and get your questions answered</p>
<a class="with-right-arrow" href="https://pytorch.org/resources">View Resources</a>
</div>
</div>
</div>
</div>