forked from pytorch/pytorch.github.io
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcpp_extension.html
800 lines (611 loc) · 43.6 KB
/
cpp_extension.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
<!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 charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>torch.utils.cpp_extension — PyTorch master documentation</title>
<link rel="canonical" href="https://pytorch.org/docs/stable/cpp_extension.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="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="index" title="Index" href="genindex.html" />
<link rel="search" title="Search" href="search.html" />
<link rel="next" title="torch.utils.data" href="data.html" />
<link rel="prev" title="torch.utils.checkpoint" href="checkpoint.html" />
<script src="_static/js/modernizr.min.js"></script>
</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>
<a href="https://pytorch.org/get-started">Get Started</a>
</li>
<li>
<a href="https://pytorch.org/features">Features</a>
</li>
<li>
<a href="https://pytorch.org/ecosystem">Ecosystem</a>
</li>
<li>
<a href="https://pytorch.org/blog/">Blog</a>
</li>
<li>
<a href="https://pytorch.org/tutorials">Tutorials</a>
</li>
<li class="active">
<a href="https://pytorch.org/docs/stable/index.html">Docs</a>
</li>
<li>
<a href="https://pytorch.org/resources">Resources</a>
</li>
<li>
<a href="https://github.com/pytorch/pytorch">Github</a>
</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='http://pytorch.org/docs/versions.html'>1.3.0a0+77c08aa ▼</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>
<div>
<a style="color:#F05732" href="https://pytorch.org/docs/stable/cpp_extension.html">
You are viewing unstable developer preview docs.
Click here to view docs for latest stable release.
</a>
</div>
<p class="caption"><span class="caption-text">Notes</span></p>
<ul>
<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/extending.html">Extending PyTorch</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/large_scale_deployments.html">Features for large-scale deployments</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/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"><span class="caption-text">Community</span></p>
<ul>
<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/governance.html">PyTorch Governance</a></li>
<li class="toctree-l1"><a class="reference internal" href="community/persons_of_interest.html">PyTorch Governance | Persons of Interest</a></li>
</ul>
<p class="caption"><span class="caption-text">Package Reference</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="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="type_info.html">Type Info</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="cuda.html">torch.cuda</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="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="nn.init.html">torch.nn.init</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="autograd.html">torch.autograd</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="distributions.html">torch.distributions</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="random.html">torch.random</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 current"><a class="current reference internal" href="#">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="dlpack.html">torch.utils.dlpack</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="onnx.html">torch.onnx</a></li>
<li class="toctree-l1"><a class="reference internal" href="__config__.html">torch.__config__</a></li>
</ul>
<p class="caption"><span class="caption-text">torchvision Reference</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="torchvision/index.html">torchvision</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.utils.cpp_extension</li>
<li class="pytorch-breadcrumbs-aside">
<a href="_sources/cpp_extension.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">
<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="torch-utils-cpp-extension">
<h1>torch.utils.cpp_extension<a class="headerlink" href="#torch-utils-cpp-extension" title="Permalink to this headline">¶</a></h1>
<dl class="function">
<dt id="torch.utils.cpp_extension.CppExtension">
<code class="sig-prename descclassname">torch.utils.cpp_extension.</code><code class="sig-name descname">CppExtension</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">sources</em>, <em class="sig-param">*args</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/utils/cpp_extension.html#CppExtension"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.utils.cpp_extension.CppExtension" title="Permalink to this definition">¶</a></dt>
<dd><p>Creates a <code class="xref py py-class docutils literal notranslate"><span class="pre">setuptools.Extension</span></code> for C++.</p>
<p>Convenience method that creates a <code class="xref py py-class docutils literal notranslate"><span class="pre">setuptools.Extension</span></code> with the
bare minimum (but often sufficient) arguments to build a C++ extension.</p>
<p>All arguments are forwarded to the <code class="xref py py-class docutils literal notranslate"><span class="pre">setuptools.Extension</span></code>
constructor.</p>
<p class="rubric">Example</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">setuptools</span> <span class="k">import</span> <span class="n">setup</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">torch.utils.cpp_extension</span> <span class="k">import</span> <span class="n">BuildExtension</span><span class="p">,</span> <span class="n">CppExtension</span>
<span class="gp">>>> </span><span class="n">setup</span><span class="p">(</span>
<span class="go"> name='extension',</span>
<span class="go"> ext_modules=[</span>
<span class="go"> CppExtension(</span>
<span class="go"> name='extension',</span>
<span class="go"> sources=['extension.cpp'],</span>
<span class="go"> extra_compile_args=['-g']),</span>
<span class="go"> ],</span>
<span class="go"> cmdclass={</span>
<span class="go"> 'build_ext': BuildExtension</span>
<span class="go"> })</span>
</pre></div>
</div>
</dd></dl>
<dl class="function">
<dt id="torch.utils.cpp_extension.CUDAExtension">
<code class="sig-prename descclassname">torch.utils.cpp_extension.</code><code class="sig-name descname">CUDAExtension</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">sources</em>, <em class="sig-param">*args</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/utils/cpp_extension.html#CUDAExtension"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.utils.cpp_extension.CUDAExtension" title="Permalink to this definition">¶</a></dt>
<dd><p>Creates a <code class="xref py py-class docutils literal notranslate"><span class="pre">setuptools.Extension</span></code> for CUDA/C++.</p>
<p>Convenience method that creates a <code class="xref py py-class docutils literal notranslate"><span class="pre">setuptools.Extension</span></code> with the
bare minimum (but often sufficient) arguments to build a CUDA/C++
extension. This includes the CUDA include path, library path and runtime
library.</p>
<p>All arguments are forwarded to the <code class="xref py py-class docutils literal notranslate"><span class="pre">setuptools.Extension</span></code>
constructor.</p>
<p class="rubric">Example</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">setuptools</span> <span class="k">import</span> <span class="n">setup</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">torch.utils.cpp_extension</span> <span class="k">import</span> <span class="n">BuildExtension</span><span class="p">,</span> <span class="n">CUDAExtension</span>
<span class="gp">>>> </span><span class="n">setup</span><span class="p">(</span>
<span class="go"> name='cuda_extension',</span>
<span class="go"> ext_modules=[</span>
<span class="go"> CUDAExtension(</span>
<span class="go"> name='cuda_extension',</span>
<span class="go"> sources=['extension.cpp', 'extension_kernel.cu'],</span>
<span class="go"> extra_compile_args={'cxx': ['-g'],</span>
<span class="go"> 'nvcc': ['-O2']})</span>
<span class="go"> ],</span>
<span class="go"> cmdclass={</span>
<span class="go"> 'build_ext': BuildExtension</span>
<span class="go"> })</span>
</pre></div>
</div>
</dd></dl>
<dl class="function">
<dt id="torch.utils.cpp_extension.BuildExtension">
<code class="sig-prename descclassname">torch.utils.cpp_extension.</code><code class="sig-name descname">BuildExtension</code><span class="sig-paren">(</span><em class="sig-param">*args</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/utils/cpp_extension.html#BuildExtension"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.utils.cpp_extension.BuildExtension" title="Permalink to this definition">¶</a></dt>
<dd><p>A custom <code class="xref py py-mod docutils literal notranslate"><span class="pre">setuptools</span></code> build extension .</p>
<p>This <code class="xref py py-class docutils literal notranslate"><span class="pre">setuptools.build_ext</span></code> subclass takes care of passing the
minimum required compiler flags (e.g. <code class="docutils literal notranslate"><span class="pre">-std=c++11</span></code>) as well as mixed
C++/CUDA compilation (and support for CUDA files in general).</p>
<p>When using <a class="reference internal" href="#torch.utils.cpp_extension.BuildExtension" title="torch.utils.cpp_extension.BuildExtension"><code class="xref py py-class docutils literal notranslate"><span class="pre">BuildExtension</span></code></a>, it is allowed to supply a dictionary
for <code class="docutils literal notranslate"><span class="pre">extra_compile_args</span></code> (rather than the usual list) that maps from
languages (<code class="docutils literal notranslate"><span class="pre">cxx</span></code> or <code class="docutils literal notranslate"><span class="pre">cuda</span></code>) to a list of additional compiler flags to
supply to the compiler. This makes it possible to supply different flags to
the C++ and CUDA compiler during mixed compilation.</p>
</dd></dl>
<dl class="function">
<dt id="torch.utils.cpp_extension.load">
<code class="sig-prename descclassname">torch.utils.cpp_extension.</code><code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">sources</em>, <em class="sig-param">extra_cflags=None</em>, <em class="sig-param">extra_cuda_cflags=None</em>, <em class="sig-param">extra_ldflags=None</em>, <em class="sig-param">extra_include_paths=None</em>, <em class="sig-param">build_directory=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">with_cuda=None</em>, <em class="sig-param">is_python_module=True</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/utils/cpp_extension.html#load"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.utils.cpp_extension.load" title="Permalink to this definition">¶</a></dt>
<dd><p>Loads a PyTorch C++ extension just-in-time (JIT).</p>
<p>To load an extension, a Ninja build file is emitted, which is used to
compile the given sources into a dynamic library. This library is
subsequently loaded into the current Python process as a module and
returned from this function, ready for use.</p>
<p>By default, the directory to which the build file is emitted and the
resulting library compiled to is <code class="docutils literal notranslate"><span class="pre"><tmp>/torch_extensions/<name></span></code>, where
<code class="docutils literal notranslate"><span class="pre"><tmp></span></code> is the temporary folder on the current platform and <code class="docutils literal notranslate"><span class="pre"><name></span></code>
the name of the extension. This location can be overridden in two ways.
First, if the <code class="docutils literal notranslate"><span class="pre">TORCH_EXTENSIONS_DIR</span></code> environment variable is set, it
replaces <code class="docutils literal notranslate"><span class="pre"><tmp>/torch_extensions</span></code> and all extensions will be compiled
into subfolders of this directory. Second, if the <code class="docutils literal notranslate"><span class="pre">build_directory</span></code>
argument to this function is supplied, it overrides the entire path, i.e.
the library will be compiled into that folder directly.</p>
<p>To compile the sources, the default system compiler (<code class="docutils literal notranslate"><span class="pre">c++</span></code>) is used,
which can be overridden by setting the <code class="docutils literal notranslate"><span class="pre">CXX</span></code> environment variable. To pass
additional arguments to the compilation process, <code class="docutils literal notranslate"><span class="pre">extra_cflags</span></code> or
<code class="docutils literal notranslate"><span class="pre">extra_ldflags</span></code> can be provided. For example, to compile your extension
with optimizations, pass <code class="docutils literal notranslate"><span class="pre">extra_cflags=['-O3']</span></code>. You can also use
<code class="docutils literal notranslate"><span class="pre">extra_cflags</span></code> to pass further include directories.</p>
<p>CUDA support with mixed compilation is provided. Simply pass CUDA source
files (<code class="docutils literal notranslate"><span class="pre">.cu</span></code> or <code class="docutils literal notranslate"><span class="pre">.cuh</span></code>) along with other sources. Such files will be
detected and compiled with nvcc rather than the C++ compiler. This includes
passing the CUDA lib64 directory as a library directory, and linking
<code class="docutils literal notranslate"><span class="pre">cudart</span></code>. You can pass additional flags to nvcc via
<code class="docutils literal notranslate"><span class="pre">extra_cuda_cflags</span></code>, just like with <code class="docutils literal notranslate"><span class="pre">extra_cflags</span></code> for C++. Various
heuristics for finding the CUDA install directory are used, which usually
work fine. If not, setting the <code class="docutils literal notranslate"><span class="pre">CUDA_HOME</span></code> environment variable is the
safest option.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>name</strong> – The name of the extension to build. This MUST be the same as the
name of the pybind11 module!</p></li>
<li><p><strong>sources</strong> – A list of relative or absolute paths to C++ source files.</p></li>
<li><p><strong>extra_cflags</strong> – optional list of compiler flags to forward to the build.</p></li>
<li><p><strong>extra_cuda_cflags</strong> – optional list of compiler flags to forward to nvcc
when building CUDA sources.</p></li>
<li><p><strong>extra_ldflags</strong> – optional list of linker flags to forward to the build.</p></li>
<li><p><strong>extra_include_paths</strong> – optional list of include directories to forward
to the build.</p></li>
<li><p><strong>build_directory</strong> – optional path to use as build workspace.</p></li>
<li><p><strong>verbose</strong> – If <code class="docutils literal notranslate"><span class="pre">True</span></code>, turns on verbose logging of load steps.</p></li>
<li><p><strong>with_cuda</strong> – Determines whether CUDA headers and libraries are added to
the build. If set to <code class="docutils literal notranslate"><span class="pre">None</span></code> (default), this value is
automatically determined based on the existence of <code class="docutils literal notranslate"><span class="pre">.cu</span></code> or
<code class="docutils literal notranslate"><span class="pre">.cuh</span></code> in <code class="docutils literal notranslate"><span class="pre">sources</span></code>. Set it to <cite>True`</cite> to force CUDA headers
and libraries to be included.</p></li>
<li><p><strong>is_python_module</strong> – If <code class="docutils literal notranslate"><span class="pre">True</span></code> (default), imports the produced shared
library as a Python module. If <code class="docutils literal notranslate"><span class="pre">False</span></code>, loads it into the process
as a plain dynamic library.</p></li>
</ul>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>If <code class="docutils literal notranslate"><span class="pre">is_python_module</span></code> is <code class="docutils literal notranslate"><span class="pre">True</span></code>, returns the loaded PyTorch
extension as a Python module. If <code class="docutils literal notranslate"><span class="pre">is_python_module</span></code> is <code class="docutils literal notranslate"><span class="pre">False</span></code>
returns nothing (the shared library is loaded into the process as a side
effect).</p>
</dd>
</dl>
<p class="rubric">Example</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">torch.utils.cpp_extension</span> <span class="k">import</span> <span class="n">load</span>
<span class="gp">>>> </span><span class="n">module</span> <span class="o">=</span> <span class="n">load</span><span class="p">(</span>
<span class="go"> name='extension',</span>
<span class="go"> sources=['extension.cpp', 'extension_kernel.cu'],</span>
<span class="go"> extra_cflags=['-O2'],</span>
<span class="go"> verbose=True)</span>
</pre></div>
</div>
</dd></dl>
<dl class="function">
<dt id="torch.utils.cpp_extension.load_inline">
<code class="sig-prename descclassname">torch.utils.cpp_extension.</code><code class="sig-name descname">load_inline</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">cpp_sources</em>, <em class="sig-param">cuda_sources=None</em>, <em class="sig-param">functions=None</em>, <em class="sig-param">extra_cflags=None</em>, <em class="sig-param">extra_cuda_cflags=None</em>, <em class="sig-param">extra_ldflags=None</em>, <em class="sig-param">extra_include_paths=None</em>, <em class="sig-param">build_directory=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">with_cuda=None</em>, <em class="sig-param">is_python_module=True</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/utils/cpp_extension.html#load_inline"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.utils.cpp_extension.load_inline" title="Permalink to this definition">¶</a></dt>
<dd><p>Loads a PyTorch C++ extension just-in-time (JIT) from string sources.</p>
<p>This function behaves exactly like <a class="reference internal" href="#torch.utils.cpp_extension.load" title="torch.utils.cpp_extension.load"><code class="xref py py-func docutils literal notranslate"><span class="pre">load()</span></code></a>, but takes its sources as
strings rather than filenames. These strings are stored to files in the
build directory, after which the behavior of <a class="reference internal" href="#torch.utils.cpp_extension.load_inline" title="torch.utils.cpp_extension.load_inline"><code class="xref py py-func docutils literal notranslate"><span class="pre">load_inline()</span></code></a> is
identical to <a class="reference internal" href="#torch.utils.cpp_extension.load" title="torch.utils.cpp_extension.load"><code class="xref py py-func docutils literal notranslate"><span class="pre">load()</span></code></a>.</p>
<p>See <a class="reference external" href="https://github.com/pytorch/pytorch/blob/master/test/test_cpp_extensions.py">the
tests</a>
for good examples of using this function.</p>
<p>Sources may omit two required parts of a typical non-inline C++ extension:
the necessary header includes, as well as the (pybind11) binding code. More
precisely, strings passed to <code class="docutils literal notranslate"><span class="pre">cpp_sources</span></code> are first concatenated into a
single <code class="docutils literal notranslate"><span class="pre">.cpp</span></code> file. This file is then prepended with <code class="docutils literal notranslate"><span class="pre">#include</span>
<span class="pre"><torch/extension.h></span></code>.</p>
<p>Furthermore, if the <code class="docutils literal notranslate"><span class="pre">functions</span></code> argument is supplied, bindings will be
automatically generated for each function specified. <code class="docutils literal notranslate"><span class="pre">functions</span></code> can
either be a list of function names, or a dictionary mapping from function
names to docstrings. If a list is given, the name of each function is used
as its docstring.</p>
<p>The sources in <code class="docutils literal notranslate"><span class="pre">cuda_sources</span></code> are concatenated into a separate <code class="docutils literal notranslate"><span class="pre">.cu</span></code>
file and prepended with <code class="docutils literal notranslate"><span class="pre">torch/types.h</span></code>, <code class="docutils literal notranslate"><span class="pre">cuda.h</span></code> and
<code class="docutils literal notranslate"><span class="pre">cuda_runtime.h</span></code> includes. The <code class="docutils literal notranslate"><span class="pre">.cpp</span></code> and <code class="docutils literal notranslate"><span class="pre">.cu</span></code> files are compiled
separately, but ultimately linked into a single library. Note that no
bindings are generated for functions in <code class="docutils literal notranslate"><span class="pre">cuda_sources</span></code> per se. To bind
to a CUDA kernel, you must create a C++ function that calls it, and either
declare or define this C++ function in one of the <code class="docutils literal notranslate"><span class="pre">cpp_sources</span></code> (and
include its name in <code class="docutils literal notranslate"><span class="pre">functions</span></code>).</p>
<p>See <a class="reference internal" href="#torch.utils.cpp_extension.load" title="torch.utils.cpp_extension.load"><code class="xref py py-func docutils literal notranslate"><span class="pre">load()</span></code></a> for a description of arguments omitted below.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>cpp_sources</strong> – A string, or list of strings, containing C++ source code.</p></li>
<li><p><strong>cuda_sources</strong> – A string, or list of strings, containing CUDA source code.</p></li>
<li><p><strong>functions</strong> – A list of function names for which to generate function
bindings. If a dictionary is given, it should map function names to
docstrings (which are otherwise just the function names).</p></li>
<li><p><strong>with_cuda</strong> – Determines whether CUDA headers and libraries are added to
the build. If set to <code class="docutils literal notranslate"><span class="pre">None</span></code> (default), this value is
automatically determined based on whether <code class="docutils literal notranslate"><span class="pre">cuda_sources</span></code> is
provided. Set it to <cite>True`</cite> to force CUDA headers
and libraries to be included.</p></li>
</ul>
</dd>
</dl>
<p class="rubric">Example</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">torch.utils.cpp_extension</span> <span class="k">import</span> <span class="n">load_inline</span>
<span class="gp">>>> </span><span class="n">source</span> <span class="o">=</span> <span class="s1">'''</span>
<span class="go">at::Tensor sin_add(at::Tensor x, at::Tensor y) {</span>
<span class="go"> return x.sin() + y.sin();</span>
<span class="go">}</span>
<span class="go">'''</span>
<span class="gp">>>> </span><span class="n">module</span> <span class="o">=</span> <span class="n">load_inline</span><span class="p">(</span><span class="n">name</span><span class="o">=</span><span class="s1">'inline_extension'</span><span class="p">,</span>
<span class="go"> cpp_sources=[source],</span>
<span class="go"> functions=['sin_add'])</span>
</pre></div>
</div>
</dd></dl>
<dl class="function">
<dt id="torch.utils.cpp_extension.include_paths">
<code class="sig-prename descclassname">torch.utils.cpp_extension.</code><code class="sig-name descname">include_paths</code><span class="sig-paren">(</span><em class="sig-param">cuda=False</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/utils/cpp_extension.html#include_paths"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.utils.cpp_extension.include_paths" title="Permalink to this definition">¶</a></dt>
<dd><p>Get the include paths required to build a C++ or CUDA extension.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>cuda</strong> – If <cite>True</cite>, includes CUDA-specific include paths.</p>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>A list of include path strings.</p>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.utils.cpp_extension.check_compiler_abi_compatibility">
<code class="sig-prename descclassname">torch.utils.cpp_extension.</code><code class="sig-name descname">check_compiler_abi_compatibility</code><span class="sig-paren">(</span><em class="sig-param">compiler</em><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/utils/cpp_extension.html#check_compiler_abi_compatibility"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.utils.cpp_extension.check_compiler_abi_compatibility" title="Permalink to this definition">¶</a></dt>
<dd><p>Verifies that the given compiler is ABI-compatible with PyTorch.</p>
<dl class="field-list simple">
<dt class="field-odd">Parameters</dt>
<dd class="field-odd"><p><strong>compiler</strong> (<a class="reference external" href="https://docs.python.org/3/library/stdtypes.html#str" title="(in Python v3.7)"><em>str</em></a>) – The compiler executable name to check (e.g. <code class="docutils literal notranslate"><span class="pre">g++</span></code>).
Must be executable in a shell process.</p>
</dd>
<dt class="field-even">Returns</dt>
<dd class="field-even"><p>False if the compiler is (likely) ABI-incompatible with PyTorch,
else True.</p>
</dd>
</dl>
</dd></dl>
<dl class="function">
<dt id="torch.utils.cpp_extension.verify_ninja_availability">
<code class="sig-prename descclassname">torch.utils.cpp_extension.</code><code class="sig-name descname">verify_ninja_availability</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="_modules/torch/utils/cpp_extension.html#verify_ninja_availability"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#torch.utils.cpp_extension.verify_ninja_availability" title="Permalink to this definition">¶</a></dt>
<dd><p>Returns <code class="docutils literal notranslate"><span class="pre">True</span></code> if the <a class="reference external" href="https://ninja-build.org/">ninja</a> build system is
available on the system.</p>
</dd></dl>
</div>
</article>
</div>
<footer>
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
<a href="data.html" class="btn btn-neutral float-right" title="torch.utils.data" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
<a href="checkpoint.html" class="btn btn-neutral" title="torch.utils.checkpoint" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
</div>
<hr>
<div role="contentinfo">
<p>
© Copyright 2019, Torch 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>
</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.utils.cpp_extension</a></li>
</ul>
</div>
</div>
</div>
</section>
</div>
<script type="text/javascript" id="documentation_options" data-url_root="./" src="_static/documentation_options.js"></script>
<script type="text/javascript" src="_static/jquery.js"></script>
<script type="text/javascript" src="_static/underscore.js"></script>
<script type="text/javascript" src="_static/doctools.js"></script>
<script type="text/javascript" src="_static/language_data.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 type="text/javascript" src="_static/js/theme.js"></script>
<script type="text/javascript">
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
<script>
(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
})(window,document,'script','https://www.google-analytics.com/analytics.js','ga');
ga('create', 'UA-90545585-1', 'auto');
ga('send', 'pageview');
</script>
<script async src="https://www.googletagmanager.com/gtag/js?id=UA-117752657-2"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'UA-117752657-2');
</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>
<footer class="site-footer">
<div class="container footer-container">
<div class="footer-logo-wrapper">
<a href="https://pytorch.org/" class="footer-logo"></a>
</div>
<div class="footer-links-wrapper">
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="https://pytorch.org/">PyTorch</a></li>
<li><a href="https://pytorch.org/get-started">Get Started</a></li>
<li><a href="https://pytorch.org/features">Features</a></li>
<li><a href="https://pytorch.org/ecosystem">Ecosystem</a></li>
<li><a href="https://pytorch.org/blog/">Blog</a></li>
<li><a href="https://pytorch.org/resources">Resources</a></li>
</ul>
</div>
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="https://pytorch.org/support">Support</a></li>
<li><a href="https://pytorch.org/tutorials">Tutorials</a></li>
<li><a href="https://pytorch.org/docs/stable/index.html">Docs</a></li>
<li><a href="https://discuss.pytorch.org" target="_blank">Discuss</a></li>
<li><a href="https://github.com/pytorch/pytorch/issues" target="_blank">Github Issues</a></li>
<li><a href="https://pytorch.slack.com" target="_blank">Slack</a></li>
<li><a href="https://github.com/pytorch/pytorch/blob/master/CONTRIBUTING.md" target="_blank">Contributing</a></li>
</ul>
</div>
<div class="footer-links-col follow-us-col">
<ul>
<li class="list-title">Follow Us</li>
<li>
<div id="mc_embed_signup">
<form
action="https://twitter.us14.list-manage.com/subscribe/post?u=75419c71fe0a935e53dfa4a3f&id=91d0dccd39"
method="post"
id="mc-embedded-subscribe-form"
name="mc-embedded-subscribe-form"
class="email-subscribe-form validate"
target="_blank"
novalidate>
<div id="mc_embed_signup_scroll" class="email-subscribe-form-fields-wrapper">
<div class="mc-field-group">
<label for="mce-EMAIL" style="display:none;">Email Address</label>
<input type="email" value="" name="EMAIL" class="required email" id="mce-EMAIL" placeholder="Email Address">
</div>
<div id="mce-responses" class="clear">
<div class="response" id="mce-error-response" style="display:none"></div>
<div class="response" id="mce-success-response" style="display:none"></div>
</div> <!-- real people should not fill this in and expect good things - do not remove this or risk form bot signups-->
<div style="position: absolute; left: -5000px;" aria-hidden="true"><input type="text" name="b_75419c71fe0a935e53dfa4a3f_91d0dccd39" tabindex="-1" value=""></div>
<div class="clear">
<input type="submit" value="" name="subscribe" id="mc-embedded-subscribe" class="button email-subscribe-button">
</div>
</div>
</form>
</div>
</li>
</ul>
<div class="footer-social-icons">
<a href="https://www.facebook.com/pytorch" target="_blank" class="facebook"></a>
<a href="https://twitter.com/pytorch" target="_blank" class="twitter"></a>
</div>
</div>
</div>
</div>
</footer>
<div class="cookie-banner-wrapper">
<div class="container">
<p class="gdpr-notice">To analyze traffic and optimize your experience, we serve cookies on this site. By clicking or navigating, you agree to allow our usage of cookies. As the current maintainers of this site, Facebook’s Cookies Policy applies. Learn more, including about available controls: <a href="https://www.facebook.com/policies/cookies/">Cookies Policy</a>.</p>
<img class="close-button" src="_static/images/pytorch-x.svg">
</div>
</div>
<!-- End Footer -->
<!-- Begin Mobile Menu -->
<div class="mobile-main-menu">
<div class="container-fluid">
<div class="container">
<div class="mobile-main-menu-header-container">
<a class="header-logo" href="https://pytorch.org/" aria-label="PyTorch"></a>
<a class="main-menu-close-button" href="#" data-behavior="close-mobile-menu"></a>
</div>
</div>
</div>
<div class="mobile-main-menu-links-container">
<div class="main-menu">
<ul>
<li>
<a href="#">Get Started</a>
</li>
<li>
<a href="#">Features</a>
</li>
<li>
<a href="#">Ecosystem</a>
</li>
<li>
<a href="https://pytorch.org/blog/">Blog</a>
</li>
<li>
<a href="https://pytorch.org/tutorials">Tutorials</a>
</li>
<li class="active">
<a href="https://pytorch.org/docs/stable/index.html">Docs</a>
</li>
<li>
<a href="https://pytorch.org/resources">Resources</a>
</li>
<li>
<a href="https://github.com/pytorch/pytorch">Github</a>
</li>
</ul>
</div>
</div>
</div>
<!-- End Mobile Menu -->
<script type="text/javascript" src="_static/js/vendor/anchor.min.js"></script>
<script type="text/javascript">
$(document).ready(function() {
mobileMenu.bind();
mobileTOC.bind();
pytorchAnchors.bind();
sideMenus.bind();
scrollToAnchor.bind();
highlightNavigation.bind();
// Add class to links that have code blocks, since we cannot create links in code blocks
$("article.pytorch-article a span.pre").each(function(e) {
$(this).closest("a").addClass("has-code");
});
})
</script>
</body>
</html>