Anurag1734/cuda-error-resolution-analysis
07
1[2 {3 "post_stream": {4 "posts": [5 {6 "id": 216214,7 "name": "A curious guy here!",8 "username": "Shisho_Sama",9 "avatar_template": "/user_avatar/discuss.pytorch.org/shisho_sama/{size}/6926_2.png",10 "created_at": "2020-07-28T05:11:54.403Z",11 "cooked": "<p>Hi everyone.<br>\nI’m in the process of converting our models from Python to C++ for production. For that it seems we need to build Pytorch from source as the prebuilt libtorch doesnt come with torch.h headers and alike !<br>\nI’m on Windows 10 1803 and have Visual Studio 2019 installed (v16.6.3). I dont have any GPUs so I didnt install CUDA! or NVTX.<br>\nI have anaconda and python 3.7 installed and all the deps are installed using the following command:</p>\n<pre><code class=\"lang-auto\">conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi\n</code></pre>\n<p>I cloned Pytorch as stated in the readme :</p>\n<pre><code class=\"lang-auto\">git clone --recursive https://github.com/pytorch/pytorch\ncd pytorch\n# if you are updating an existing checkout\ngit submodule sync\ngit submodule update --init --recursive\n</code></pre>\n<p>But upon executing these commands :</p>\n<pre><code class=\"lang-auto\">:: [Optional] If you want to build with VS 2019 generator, please change the value in the next line to `Visual Studio 16 2019`.\n:: Note: This value is useless if Ninja is detected. However, you can force that by using `set USE_NINJA=OFF`.\nset CMAKE_GENERATOR=Visual Studio 16 2019\n\n:: Read the content in the previous section carefully before you proceed.\n:: [Optional] If you want to override the underlying toolset used by Ninja and Visual Studio with CUDA, please run the following script block.\n:: \"Visual Studio 2017 Developer Command Prompt\" will be run automatically.\n:: Make sure you have CMake >= 3.12 before you do this when you use the Visual Studio generator.\nset CMAKE_GENERATOR_TOOLSET_VERSION=14.11\nset DISTUTILS_USE_SDK=1\nfor /f \"usebackq tokens=*\" %i in (`\"%ProgramFiles(x86)%\\Microsoft Visual Studio\\Installer\\vswhere.exe\" -version [15^,16^) -products * -latest -property installationPath`) do call \"%i\\VC\\Auxiliary\\Build\\vcvarsall.bat\" x64 -vcvars_ver=%CMAKE_GENERATOR_TOOLSET_VERSION%\n\n:: [Optional] If you want to override the cuda host compiler\nset CUDAHOSTCXX=C:\\Program Files (x86)\\Microsoft Visual Studio\\2017\\Enterprise\\VC\\Tools\\MSVC\\14.11.25503\\bin\\HostX64\\x64\\cl.exe\n\npython setup.py install\n</code></pre>\n<p>I face these errors:</p>\n<pre><code class=\"lang-auto\">D:\\Codes\\pytorch>python setup.py install\nBuilding wheel torch-1.7.0a0+8376284\n-- Building version 1.7.0a0+8376284\ncmake -GNinja -DBUILD_PYTHON=True -DBUILD_TEST=True -DCMAKE_BUILD_TYPE=Release -DCMAKE_GENERATOR=Visual Studio 16 2019 -DCMAKE_GENERATOR_TOOLSET_VERSION=14.11 -DCMAKE_INSTALL_PREFIX=D:\\Codes\\pytorch\\torch -DCMAKE_PREFIX_PATH=C:\\Users\\User\\Anaconda3\\Lib\\site-packages -DNUMPY_INCLUDE_DIR=C:\\Users\\User\\Anaconda3\\Lib\\site-packages\\numpy\\core\\include -DPYTHON_EXECUTABLE=C:\\Users\\User\\Anaconda3\\python.exe -DPYTHON_INCLUDE_DIR=C:\\Users\\User\\Anaconda3\\include -DPYTHON_LIBRARY=C:\\Users\\User\\Anaconda3/libs/python37.lib -DTORCH_BUILD_VERSION=1.7.0a0+8376284 -DUSE_NUMPY=True D:\\Codes\\pytorch\nCMake Error: Error: generator : Ninja\nDoes not match the generator used previously: Visual Studio 16 2019\nEither remove the CMakeCache.txt file and CMakeFiles directory or choose a different binary directory.\nTraceback (most recent call last):\n File \"setup.py\", line 737, in <module>\n build_deps()\n File \"setup.py\", line 321, in build_deps\n cmake=cmake)\n File \"D:\\Codes\\pytorch\\tools\\build_pytorch_libs.py\", line 59, in build_caffe2\n rerun_cmake)\n File \"D:\\Codes\\pytorch\\tools\\setup_helpers\\cmake.py\", line 329, in generate\n self.run(args, env=my_env)\n File \"D:\\Codes\\pytorch\\tools\\setup_helpers\\cmake.py\", line 141, in run\n check_call(command, cwd=self.build_dir, env=env)\n File \"C:\\Users\\User\\Anaconda3\\Lib\\subprocess.py\", line 347, in check_call\n raise CalledProcessError(retcode, cmd)\nsubprocess.CalledProcessError: Command '['cmake', '-GNinja', '-DBUILD_PYTHON=True', '-DBUILD_TEST=True', '-DCMAKE_BUILD_TYPE=Release', '-DCMAKE_GENERATOR=Visual Studio 16 2019', '-DCMAKE_GENERATOR_TOOLSET_VERSION=14.11', '-DCMAKE_INSTALL_PREFIX=D:\\\\Codes\\\\pytorch\\\\torch', '-DCMAKE_PREFIX_PATH=C:\\\\Users\\\\User\\\\Anaconda3\\\\Lib\\\\site-packages', '-DNUMPY_INCLUDE_DIR=C:\\\\Users\\\\User\\\\Anaconda3\\\\Lib\\\\site-packages\\\\numpy\\\\core\\\\include', '-DPYTHON_EXECUTABLE=C:\\\\Users\\\\User\\\\Anaconda3\\\\python.exe', '-DPYTHON_INCLUDE_DIR=C:\\\\Users\\\\User\\\\Anaconda3\\\\include', '-DPYTHON_LIBRARY=C:\\\\Users\\\\User\\\\Anaconda3/libs/python37.lib', '-DTORCH_BUILD_VERSION=1.7.0a0+8376284', '-DUSE_NUMPY=True', 'D:\\\\Codes\\\\pytorch']' returned non-zero exit status 1.\n</code></pre>\n<p>What am I doing wrong?<br>\nAny help is greatly appreciated</p>",12 "post_number": 1,13 "post_type": 1,14 "posts_count": 3,15 "updated_at": "2020-07-28T05:18:53.669Z",16 "reply_count": 0,17 "reply_to_post_number": null,18 "quote_count": 0,19 "incoming_link_count": 8724,20 "reads": 48,21 "readers_count": 47,22 "score": 43609.6,23 "yours": false,24 "topic_id": 90769,25 "topic_slug": "is-building-from-source-for-windows-up-to-date-subprocess-calledprocesserror-command-cmake-returned-non-zero-exit-status-1-error",26 "display_username": "A curious guy here!",27 "primary_group_name": null,28 "flair_name": null,29 "flair_url": null,30 "flair_bg_color": null,31 "flair_color": null,32 "flair_group_id": null,33 "badges_granted": [],34 "version": 2,35 "can_edit": false,36 "can_delete": false,37 "can_recover": false,38 "can_see_hidden_post": false,39 "can_wiki": false,40 "link_counts": [41 {42 "url": "https://discuss.pytorch.org/t/the-ordinal-961-could-not-be-located-in-the-dynamic-link-library/90791",43 "internal": true,44 "reflection": true,45 "title": "The ordinal 961 could not be located in the dynamic link library",46 "clicks": 347 }48 ],49 "read": true,50 "user_title": "",51 "bookmarked": false,52 "actions_summary": [],53 "moderator": false,54 "admin": false,55 "staff": false,56 "user_id": 7863,57 "hidden": false,58 "trust_level": 2,59 "deleted_at": null,60 "user_deleted": false,61 "edit_reason": null,62 "can_view_edit_history": true,63 "wiki": false,64 "post_url": "/t/is-building-from-source-for-windows-up-to-date-subprocess-calledprocesserror-command-cmake-returned-non-zero-exit-status-1-error/90769/1",65 "can_accept_answer": false,66 "can_unaccept_answer": false,67 "accepted_answer": false,68 "topic_accepted_answer": true,69 "can_vote": false70 },71 {72 "id": 216235,73 "name": "A curious guy here!",74 "username": "Shisho_Sama",75 "avatar_template": "/user_avatar/discuss.pytorch.org/shisho_sama/{size}/6926_2.png",76 "created_at": "2020-07-28T07:02:08.729Z",77 "cooked": "<p>Looking at this <a href=\"https://discuss.pytorch.org/t/pytorch-build-from-source-on-windows/40288/9\">link</a> everything built successfully! without even the need for any of these :</p>\n<pre><code class=\"lang-auto\">:: [Optional] If you want to build with VS 2019 generator, please change the value in the next line to `Visual Studio 16 2019`.\n:: Note: This value is useless if Ninja is detected. However, you can force that by using `set USE_NINJA=OFF`.\nset CMAKE_GENERATOR=Visual Studio 16 2019\n\n:: Read the content in the previous section carefully before you proceed.\n:: [Optional] If you want to override the underlying toolset used by Ninja and Visual Studio with CUDA, please run the following script block.\n:: \"Visual Studio 2017 Developer Command Prompt\" will be run automatically.\n:: Make sure you have CMake >= 3.12 before you do this when you use the Visual Studio generator.\nset CMAKE_GENERATOR_TOOLSET_VERSION=14.11\nset DISTUTILS_USE_SDK=1\nfor /f \"usebackq tokens=*\" %i in (`\"%ProgramFiles(x86)%\\Microsoft Visual Studio\\Installer\\vswhere.exe\" -version [15^,16^) -products * -latest -property installationPath`) do call \"%i\\VC\\Auxiliary\\Build\\vcvarsall.bat\" x64 -vcvars_ver=%CMAKE_GENERATOR_TOOLSET_VERSION%\n\n:: [Optional] If you want to override the cuda host compiler\nset CUDAHOSTCXX=C:\\Program Files (x86)\\Microsoft Visual Studio\\2017\\Enterprise\\VC\\Tools\\MSVC\\14.11.25503\\bin\\HostX64\\x64\\cl.exe\n\npython setup.py install\n</code></pre>\n<p>I just ran :</p>\n<pre><code class=\"lang-auto\">python setup.py build --cmake\npython setup.py install\n</code></pre>\n<p>and everything got built!<br>\nHowever, after building everything, I cant seem to be able to find <torch/torch.h> ! I read it on this issue <a href=\"https://github.com/pytorch/pytorch/issues/5964#issuecomment-375698855\" rel=\"nofollow noopener\">here</a> that building from source will make that but apparently this is not the case here!</p>\n<p>Any help is greatly appreciated</p>",78 "post_number": 2,79 "post_type": 1,80 "posts_count": 3,81 "updated_at": "2020-07-28T07:02:08.729Z",82 "reply_count": 1,83 "reply_to_post_number": null,84 "quote_count": 0,85 "incoming_link_count": 124,86 "reads": 42,87 "readers_count": 41,88 "score": 633.4,89 "yours": false,90 "topic_id": 90769,91 "topic_slug": "is-building-from-source-for-windows-up-to-date-subprocess-calledprocesserror-command-cmake-returned-non-zero-exit-status-1-error",92 "display_username": "A curious guy here!",93 "primary_group_name": null,94 "flair_name": null,95 "flair_url": null,96 "flair_bg_color": null,97 "flair_color": null,98 "flair_group_id": null,99 "badges_granted": [],100 "version": 1,101 "can_edit": false,102 "can_delete": false,103 "can_recover": false,104 "can_see_hidden_post": false,105 "can_wiki": false,106 "link_counts": [107 {108 "url": "https://discuss.pytorch.org/t/pytorch-build-from-source-on-windows/40288/9",109 "internal": true,110 "reflection": false,111 "title": "PyTorch build from source on Windows",112 "clicks": 206113 },114 {115 "url": "https://github.com/pytorch/pytorch/issues/5964#issuecomment-375698855",116 "internal": false,117 "reflection": false,118 "title": "fatal error: torch/torch.h: No such file or directory · Issue #5964 · pytorch/pytorch · GitHub",119 "clicks": 11120 }121 ],122 "read": true,123 "user_title": "",124 "bookmarked": false,125 "actions_summary": [],126 "moderator": false,127 "admin": false,128 "staff": false,129 "user_id": 7863,130 "hidden": false,131 "trust_level": 2,132 "deleted_at": null,133 "user_deleted": false,134 "edit_reason": null,135 "can_view_edit_history": true,136 "wiki": false,137 "post_url": "/t/is-building-from-source-for-windows-up-to-date-subprocess-calledprocesserror-command-cmake-returned-non-zero-exit-status-1-error/90769/2",138 "can_accept_answer": false,139 "can_unaccept_answer": false,140 "accepted_answer": false,141 "topic_accepted_answer": true142 },143 {144 "id": 216266,145 "name": "A curious guy here!",146 "username": "Shisho_Sama",147 "avatar_template": "/user_avatar/discuss.pytorch.org/shisho_sama/{size}/6926_2.png",148 "created_at": "2020-07-28T08:45:34.710Z",149 "cooked": "<p>OK, Here are my findings.<br>\nFirst of all <code>include <torch/torch.h></code> doesnt need building from source. its just in the a different path that is,<br>\nin include directories, one need to add both :<br>\n<code>$(SolutionDir)Dependencies\\libtorch-debug-latest\\libtorch\\include; $(SolutionDir)Dependencies\\libtorch-debug-latest\\libtorch\\include\\torch\\csrc\\api\\include;</code><br>\nand voila you are good to go!</p>",150 "post_number": 3,151 "post_type": 1,152 "posts_count": 3,153 "updated_at": "2020-07-28T08:45:36.209Z",154 "reply_count": 0,155 "reply_to_post_number": 2,156 "quote_count": 0,157 "incoming_link_count": 131,158 "reads": 38,159 "readers_count": 37,160 "score": 662.6,161 "yours": false,162 "topic_id": 90769,163 "topic_slug": "is-building-from-source-for-windows-up-to-date-subprocess-calledprocesserror-command-cmake-returned-non-zero-exit-status-1-error",164 "display_username": "A curious guy here!",165 "primary_group_name": null,166 "flair_name": null,167 "flair_url": null,168 "flair_bg_color": null,169 "flair_color": null,170 "flair_group_id": null,171 "badges_granted": [],172 "version": 1,173 "can_edit": false,174 "can_delete": false,175 "can_recover": false,176 "can_see_hidden_post": false,177 "can_wiki": false,178 "read": true,179 "user_title": "",180 "reply_to_user": {181 "id": 7863,182 "username": "Shisho_Sama",183 "name": "A curious guy here!",184 "avatar_template": "/user_avatar/discuss.pytorch.org/shisho_sama/{size}/6926_2.png"185 },186 "bookmarked": false,187 "actions_summary": [],188 "moderator": false,189 "admin": false,190 "staff": false,191 "user_id": 7863,192 "hidden": false,193 "trust_level": 2,194 "deleted_at": null,195 "user_deleted": false,196 "edit_reason": null,197 "can_view_edit_history": true,198 "wiki": false,199 "post_url": "/t/is-building-from-source-for-windows-up-to-date-subprocess-calledprocesserror-command-cmake-returned-non-zero-exit-status-1-error/90769/3",200 "can_accept_answer": false,201 "can_unaccept_answer": false,202 "accepted_answer": true,203 "topic_accepted_answer": true204 }205 ],206 "stream": [207 216214,208 216235,209 216266210 ]211 },212 "timeline_lookup": [213 [214 1,215 1916216 ]217 ],218 "suggested_topics": [219 {220 "fancy_title": "How am i supposed to get old versions of torch on windows x64 arhitecture?",221 "id": 222813,222 "title": "How am i supposed to get old versions of torch on windows x64 arhitecture?",223 "slug": "how-am-i-supposed-to-get-old-versions-of-torch-on-windows-x64-arhitecture",224 "posts_count": 3,225 "reply_count": 1,226 "highest_post_number": 3,227 "image_url": null,228 "created_at": "2025-09-02T00:37:13.338Z",229 "last_posted_at": "2025-09-05T09:36:12.011Z",230 "bumped": true,231 "bumped_at": "2025-09-05T09:36:12.011Z",232 "archetype": "regular",233 "unseen": false,234 "pinned": false,235 "unpinned": null,236 "visible": true,237 "closed": false,238 "archived": false,239 "bookmarked": null,240 "liked": null,241 "tags_descriptions": {},242 "like_count": 0,243 "views": 35,244 "category_id": 26,245 "featured_link": null,246 "has_accepted_answer": false,247 "posters": [248 {249 "extras": "latest",250 "description": "Original Poster, Most Recent Poster",251 "user": {252 "id": 85715,253 "username": "Alex_Eugen_Botez",254 "name": "Alex Eugen Botez",255 "avatar_template": "/user_avatar/discuss.pytorch.org/alex_eugen_botez/{size}/75235_2.png",256 "trust_level": 0257 }258 },259 {260 "extras": null,261 "description": "Frequent Poster",262 "user": {263 "id": 3534,264 "username": "ptrblck",265 "name": "",266 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",267 "admin": true,268 "moderator": true,269 "trust_level": 2270 }271 }272 ]273 },274 {275 "fancy_title": "Identifying if gpu cores are used",276 "id": 216872,277 "title": "Identifying if gpu cores are used",278 "slug": "identifying-if-gpu-cores-are-used",279 "posts_count": 3,280 "reply_count": 0,281 "highest_post_number": 4,282 "image_url": "https://discuss.pytorch.org/uploads/default/optimized/3X/e/8/e89b25fc8d05172dff54f38fd989c9bea1b386ab_2_1024x545.png",283 "created_at": "2025-02-19T07:45:24.715Z",284 "last_posted_at": "2025-05-06T13:32:31.627Z",285 "bumped": true,286 "bumped_at": "2025-05-06T13:32:31.627Z",287 "archetype": "regular",288 "unseen": false,289 "pinned": false,290 "unpinned": null,291 "visible": true,292 "closed": false,293 "archived": false,294 "bookmarked": null,295 "liked": null,296 "tags_descriptions": {},297 "like_count": 0,298 "views": 204,299 "category_id": 26,300 "featured_link": null,301 "has_accepted_answer": false,302 "posters": [303 {304 "extras": null,305 "description": "Original Poster",306 "user": {307 "id": 71824,308 "username": "Stephane_Ancelot",309 "name": "Stéphane Ancelot",310 "avatar_template": "/user_avatar/discuss.pytorch.org/stephane_ancelot/{size}/61201_2.png",311 "trust_level": 1312 }313 },314 {315 "extras": null,316 "description": "Frequent Poster",317 "user": {318 "id": 3534,319 "username": "ptrblck",320 "name": "",321 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",322 "admin": true,323 "moderator": true,324 "trust_level": 2325 }326 },327 {328 "extras": "latest",329 "description": "Most Recent Poster",330 "user": {331 "id": 84077,332 "username": "python",333 "name": null,334 "avatar_template": "/letter_avatar_proxy/v4/letter/p/ee7513/{size}.png",335 "trust_level": 1336 }337 }338 ]339 },340 {341 "fancy_title": "TorchCodec wheels",342 "id": 218578,343 "title": "TorchCodec wheels",344 "slug": "torchcodec-wheels",345 "posts_count": 3,346 "reply_count": 0,347 "highest_post_number": 3,348 "image_url": null,349 "created_at": "2025-04-03T12:05:19.354Z",350 "last_posted_at": "2025-10-23T11:22:55.349Z",351 "bumped": true,352 "bumped_at": "2025-10-23T11:22:55.349Z",353 "archetype": "regular",354 "unseen": false,355 "pinned": false,356 "unpinned": null,357 "visible": true,358 "closed": false,359 "archived": false,360 "bookmarked": null,361 "liked": null,362 "tags_descriptions": {},363 "like_count": 0,364 "views": 431,365 "category_id": 26,366 "featured_link": null,367 "has_accepted_answer": false,368 "posters": [369 {370 "extras": null,371 "description": "Original Poster",372 "user": {373 "id": 83616,374 "username": "richbill",375 "name": "",376 "avatar_template": "/letter_avatar_proxy/v4/letter/r/4af34b/{size}.png",377 "trust_level": 1378 }379 },380 {381 "extras": null,382 "description": "Frequent Poster",383 "user": {384 "id": 84077,385 "username": "python",386 "name": null,387 "avatar_template": "/letter_avatar_proxy/v4/letter/p/ee7513/{size}.png",388 "trust_level": 1389 }390 },391 {392 "extras": "latest",393 "description": "Most Recent Poster",394 "user": {395 "id": 86261,396 "username": "marco_adamo",397 "name": "marco adamo",398 "avatar_template": "/user_avatar/discuss.pytorch.org/marco_adamo/{size}/78511_2.png",399 "trust_level": 0400 }401 }402 ]403 },404 {405 "fancy_title": "How to use torch with CUDA 12.9",406 "id": 220827,407 "title": "How to use torch with CUDA 12.9",408 "slug": "how-to-use-torch-with-cuda-12-9",409 "posts_count": 5,410 "reply_count": 2,411 "highest_post_number": 7,412 "image_url": null,413 "created_at": "2025-06-15T07:08:55.857Z",414 "last_posted_at": "2025-07-18T22:41:59.447Z",415 "bumped": true,416 "bumped_at": "2025-07-18T22:41:59.447Z",417 "archetype": "regular",418 "unseen": false,419 "pinned": false,420 "unpinned": null,421 "visible": true,422 "closed": false,423 "archived": false,424 "bookmarked": null,425 "liked": null,426 "tags_descriptions": {},427 "like_count": 3,428 "views": 7528,429 "category_id": 26,430 "featured_link": null,431 "has_accepted_answer": false,432 "posters": [433 {434 "extras": null,435 "description": "Original Poster",436 "user": {437 "id": 84711,438 "username": "Ahmed_Mhmouad",439 "name": "Ahmed Mhmouad",440 "avatar_template": "/user_avatar/discuss.pytorch.org/ahmed_mhmouad/{size}/77381_2.png",441 "trust_level": 0442 }443 },444 {445 "extras": null,446 "description": "Frequent Poster",447 "user": {448 "id": 3534,449 "username": "ptrblck",450 "name": "",451 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",452 "admin": true,453 "moderator": true,454 "trust_level": 2455 }456 },457 {458 "extras": null,459 "description": "Frequent Poster",460 "user": {461 "id": 85105,462 "username": "yogesh_jog",463 "name": "yogesh jog",464 "avatar_template": "/user_avatar/discuss.pytorch.org/yogesh_jog/{size}/77689_2.png",465 "trust_level": 0466 }467 },468 {469 "extras": "latest",470 "description": "Most Recent Poster",471 "user": {472 "id": 85150,473 "username": "HipHop_Delhi",474 "name": "HipHop Delhi",475 "avatar_template": "/user_avatar/discuss.pytorch.org/hiphop_delhi/{size}/77724_2.png",476 "trust_level": 0477 }478 }479 ]480 },481 {482 "fancy_title": "PyTorch scripts not on PATH",483 "id": 222873,484 "title": "PyTorch scripts not on PATH",485 "slug": "pytorch-scripts-not-on-path",486 "posts_count": 2,487 "reply_count": 0,488 "highest_post_number": 2,489 "image_url": null,490 "created_at": "2025-09-05T04:05:06.932Z",491 "last_posted_at": "2025-09-30T15:08:05.003Z",492 "bumped": true,493 "bumped_at": "2025-09-30T15:08:05.003Z",494 "archetype": "regular",495 "unseen": false,496 "pinned": false,497 "unpinned": null,498 "visible": true,499 "closed": false,500 "archived": false,501 "bookmarked": null,502 "liked": null,503 "tags_descriptions": {},504 "like_count": 0,505 "views": 97,506 "category_id": 26,507 "featured_link": null,508 "has_accepted_answer": false,509 "posters": [510 {511 "extras": null,512 "description": "Original Poster",513 "user": {514 "id": 85749,515 "username": "lfd.arch",516 "name": "Luis Donoso",517 "avatar_template": "/user_avatar/discuss.pytorch.org/lfd.arch/{size}/78141_2.png",518 "trust_level": 0519 }520 },521 {522 "extras": "latest",523 "description": "Most Recent Poster",524 "user": {525 "id": 15730,526 "username": "charan_Vjy",527 "name": "Charan Vjy",528 "avatar_template": "/user_avatar/discuss.pytorch.org/charan_vjy/{size}/13238_2.png",529 "trust_level": 2530 }531 }532 ]533 }534 ],535 "tags_descriptions": {},536 "fancy_title": "Is building from source for windows up to date? (subprocess.CalledProcessError: Command ‘[‘cmake’, ..] returned non-zero exit status 1 error)",537 "id": 90769,538 "title": "Is building from source for windows up to date? (subprocess.CalledProcessError: Command '['cmake', ..] returned non-zero exit status 1 error)",539 "posts_count": 3,540 "created_at": "2020-07-28T05:11:54.348Z",541 "views": 10738,542 "reply_count": 1,543 "like_count": 0,544 "last_posted_at": "2020-07-28T08:45:34.710Z",545 "visible": true,546 "closed": false,547 "archived": false,548 "has_summary": false,549 "archetype": "regular",550 "slug": "is-building-from-source-for-windows-up-to-date-subprocess-calledprocesserror-command-cmake-returned-non-zero-exit-status-1-error",551 "category_id": 26,552 "word_count": 964,553 "deleted_at": null,554 "user_id": 7863,555 "featured_link": null,556 "pinned_globally": false,557 "pinned_at": null,558 "pinned_until": null,559 "image_url": null,560 "slow_mode_seconds": 0,561 "draft": null,562 "draft_key": "topic_90769",563 "draft_sequence": null,564 "unpinned": null,565 "pinned": false,566 "current_post_number": 1,567 "highest_post_number": 3,568 "deleted_by": null,569 "actions_summary": [570 {571 "id": 4,572 "count": 0,573 "hidden": false,574 "can_act": false575 },576 {577 "id": 8,578 "count": 0,579 "hidden": false,580 "can_act": false581 },582 {583 "id": 10,584 "count": 0,585 "hidden": false,586 "can_act": false587 },588 {589 "id": 7,590 "count": 0,591 "hidden": false,592 "can_act": false593 }594 ],595 "chunk_size": 20,596 "bookmarked": false,597 "topic_timer": null,598 "message_bus_last_id": 0,599 "participant_count": 1,600 "show_read_indicator": false,601 "thumbnails": null,602 "slow_mode_enabled_until": null,603 "accepted_answer": {604 "post_number": 3,605 "username": "Shisho_Sama",606 "name": "A curious guy here!",607 "excerpt": "OK, Here are my findings. \nFirst of all include <torch/torch.h> doesnt need building from source. its just in the a different path that is, \nin include directories, one need to add both : \n$(SolutionDir)Dependencies\\libtorch-debug-latest\\libtorch\\include; $(SolutionDir)Dependencies\\libtorch-debug-la…"608 },609 "can_vote": false,610 "vote_count": 0,611 "user_voted": false,612 "discourse_zendesk_plugin_zendesk_id": null,613 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",614 "details": {615 "can_edit": false,616 "notification_level": 1,617 "participants": [618 {619 "id": 7863,620 "username": "Shisho_Sama",621 "name": "A curious guy here!",622 "avatar_template": "/user_avatar/discuss.pytorch.org/shisho_sama/{size}/6926_2.png",623 "post_count": 3,624 "primary_group_name": null,625 "flair_name": null,626 "flair_url": null,627 "flair_color": null,628 "flair_bg_color": null,629 "flair_group_id": null,630 "trust_level": 2631 }632 ],633 "created_by": {634 "id": 7863,635 "username": "Shisho_Sama",636 "name": "A curious guy here!",637 "avatar_template": "/user_avatar/discuss.pytorch.org/shisho_sama/{size}/6926_2.png"638 },639 "last_poster": {640 "id": 7863,641 "username": "Shisho_Sama",642 "name": "A curious guy here!",643 "avatar_template": "/user_avatar/discuss.pytorch.org/shisho_sama/{size}/6926_2.png"644 },645 "links": [646 {647 "url": "https://discuss.pytorch.org/t/pytorch-build-from-source-on-windows/40288/9",648 "title": "PyTorch build from source on Windows",649 "internal": true,650 "attachment": false,651 "reflection": false,652 "clicks": 206,653 "user_id": 7863,654 "domain": "discuss.pytorch.org",655 "root_domain": "pytorch.org"656 },657 {658 "url": "https://github.com/pytorch/pytorch/issues/5964#issuecomment-375698855",659 "title": "fatal error: torch/torch.h: No such file or directory · Issue #5964 · pytorch/pytorch · GitHub",660 "internal": false,661 "attachment": false,662 "reflection": false,663 "clicks": 11,664 "user_id": 7863,665 "domain": "github.com",666 "root_domain": "github.com"667 },668 {669 "url": "https://discuss.pytorch.org/t/the-ordinal-961-could-not-be-located-in-the-dynamic-link-library/90791",670 "title": "The ordinal 961 could not be located in the dynamic link library",671 "internal": true,672 "attachment": false,673 "reflection": true,674 "clicks": 3,675 "user_id": 7863,676 "domain": "discuss.pytorch.org",677 "root_domain": "pytorch.org"678 }679 ]680 },681 "bookmarks": []682 },683 {684 "post_stream": {685 "posts": [686 {687 "id": 212502,688 "name": "Florian Dietz",689 "username": "Florian_Dietz",690 "avatar_template": "/user_avatar/discuss.pytorch.org/florian_dietz/{size}/11502_2.png",691 "created_at": "2020-07-15T08:58:30.731Z",692 "cooked": "<p>I am working on an unusual research project and need to modify some gradients in-place.</p>\n<p>This worked fine so far, but I just updated to a newer version of Pytorch, and now I am getting this error:</p>\n<p>“RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.FloatTensor [762, 5]], which is output 0 of TBackward, is at version 5; expected version 2 instead. Hint: the backtrace further above shows the operation that failed to compute its gradient. The variable in question was changed in there or anywhere later. Good luck!”</p>\n<p>I get that this is an error for almost every usecase, but I actually want to do this on purpose.</p>\n<p>Is there a way to turn this error off?</p>\n<p>Alternatively, is there another way to modify gradients that is legal / preferred?</p>\n<p>(My situation: I’m running a network where the architecture is not fixed but changes based on intermediate results. To backpropagate through this safely, I break the network into smaller parts, and manually retrieve and re-insert the gradients at the appropriate tensors.)</p>",693 "post_number": 1,694 "post_type": 1,695 "posts_count": 8,696 "updated_at": "2020-07-15T08:58:30.731Z",697 "reply_count": 0,698 "reply_to_post_number": null,699 "quote_count": 0,700 "incoming_link_count": 2276,701 "reads": 120,702 "readers_count": 119,703 "score": 11404.0,704 "yours": false,705 "topic_id": 89290,706 "topic_slug": "how-to-modify-gradients-in-place",707 "display_username": "Florian Dietz",708 "primary_group_name": null,709 "flair_name": null,710 "flair_url": null,711 "flair_bg_color": null,712 "flair_color": null,713 "flair_group_id": null,714 "badges_granted": [],715 "version": 1,716 "can_edit": false,717 "can_delete": false,718 "can_recover": false,719 "can_see_hidden_post": false,720 "can_wiki": false,721 "read": true,722 "user_title": null,723 "bookmarked": false,724 "actions_summary": [725 {726 "id": 2,727 "count": 1728 }729 ],730 "moderator": false,731 "admin": false,732 "staff": false,733 "user_id": 18090,734 "hidden": false,735 "trust_level": 1,736 "deleted_at": null,737 "user_deleted": false,738 "edit_reason": null,739 "can_view_edit_history": true,740 "wiki": false,741 "post_url": "/t/how-to-modify-gradients-in-place/89290/1",742 "can_accept_answer": false,743 "can_unaccept_answer": false,744 "accepted_answer": false,745 "topic_accepted_answer": null,746 "can_vote": false747 },748 {749 "id": 212543,750 "name": "",751 "username": "ptrblck",752 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",753 "created_at": "2020-07-15T10:38:51.329Z",754 "cooked": "<p>Would your workflow work with hooks (via <code>tensor.register_hook()</code>) to manipulate the gradients?<br>\nIf not, could you try to wrap the gradient manipulations into a <code>with torch.no_grad()</code> block, if that’s not already the case?</p>",755 "post_number": 2,756 "post_type": 1,757 "posts_count": 8,758 "updated_at": "2020-07-15T11:28:03.838Z",759 "reply_count": 1,760 "reply_to_post_number": null,761 "quote_count": 0,762 "incoming_link_count": 14,763 "reads": 115,764 "readers_count": 114,765 "score": 98.0,766 "yours": false,767 "topic_id": 89290,768 "topic_slug": "how-to-modify-gradients-in-place",769 "display_username": "",770 "primary_group_name": null,771 "flair_name": null,772 "flair_url": null,773 "flair_bg_color": null,774 "flair_color": null,775 "flair_group_id": null,776 "badges_granted": [],777 "version": 1,778 "can_edit": false,779 "can_delete": false,780 "can_recover": false,781 "can_see_hidden_post": false,782 "can_wiki": false,783 "read": true,784 "user_title": "",785 "bookmarked": false,786 "actions_summary": [],787 "moderator": true,788 "admin": true,789 "staff": true,790 "user_id": 3534,791 "hidden": false,792 "trust_level": 2,793 "deleted_at": null,794 "user_deleted": false,795 "edit_reason": null,796 "can_view_edit_history": true,797 "wiki": false,798 "post_url": "/t/how-to-modify-gradients-in-place/89290/2",799 "can_accept_answer": false,800 "can_unaccept_answer": false,801 "accepted_answer": false,802 "topic_accepted_answer": null803 },804 {805 "id": 212561,806 "name": "Florian Dietz",807 "username": "Florian_Dietz",808 "avatar_template": "/user_avatar/discuss.pytorch.org/florian_dietz/{size}/11502_2.png",809 "created_at": "2020-07-15T11:32:07.152Z",810 "cooked": "<p><code>tensor.register_hook()</code> would have worked, but I would have to refactor everything. It would take days, and I might lose some functionality, because the system I already built is more flexible for my purposes. (Hooks get executed immediately, but I build a prioritized queue of calculations that still need to be performed)</p>\n<p>I’m not sure I understand how <code>torch.no_grad()</code> would help. I am performing normal gradient calculations through backpropagation. Then I apply modifications to the generated gradients.</p>",811 "post_number": 3,812 "post_type": 1,813 "posts_count": 8,814 "updated_at": "2020-07-15T11:32:07.152Z",815 "reply_count": 1,816 "reply_to_post_number": 2,817 "quote_count": 0,818 "incoming_link_count": 14,819 "reads": 115,820 "readers_count": 114,821 "score": 98.0,822 "yours": false,823 "topic_id": 89290,824 "topic_slug": "how-to-modify-gradients-in-place",825 "display_username": "Florian Dietz",826 "primary_group_name": null,827 "flair_name": null,828 "flair_url": null,829 "flair_bg_color": null,830 "flair_color": null,831 "flair_group_id": null,832 "badges_granted": [],833 "version": 1,834 "can_edit": false,835 "can_delete": false,836 "can_recover": false,837 "can_see_hidden_post": false,838 "can_wiki": false,839 "read": true,840 "user_title": null,841 "reply_to_user": {842 "id": 3534,843 "username": "ptrblck",844 "name": "",845 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png"846 },847 "bookmarked": false,848 "actions_summary": [],849 "moderator": false,850 "admin": false,851 "staff": false,852 "user_id": 18090,853 "hidden": false,854 "trust_level": 1,855 "deleted_at": null,856 "user_deleted": false,857 "edit_reason": null,858 "can_view_edit_history": true,859 "wiki": false,860 "post_url": "/t/how-to-modify-gradients-in-place/89290/3",861 "can_accept_answer": false,862 "can_unaccept_answer": false,863 "accepted_answer": false,864 "topic_accepted_answer": null865 },866 {867 "id": 212566,868 "name": "Thomas V",869 "username": "tom",870 "avatar_template": "/user_avatar/discuss.pytorch.org/tom/{size}/3162_2.png",871 "created_at": "2020-07-15T11:42:28.471Z",872 "cooked": "<p>Note that</p>\n<ol>\n<li>the error message comes from modifying something in-place that has been used in the forward and is needed for the backward. You might check the backtrace mentioned here: (Note that it typically is in C++ stderr and not shown in Jupyter unless you use nightlies when it is in Python stderr.)</li>\n</ol>\n<blockquote>\n<p>Hint: the backtrace further above shows the operation that failed to compute its gradient.</p>\n</blockquote>\n<ol start=\"2\">\n<li>If for the type of thing you are doing, I would probably not use <code>backward</code> for the intermediates but use <code>torch.autograd.grad</code>. Basically, <code>backward</code> is best thought of as existing for the very narrow case of “compute gradients then fed into the optimizer”.</li>\n</ol>\n<p>Best regards</p>\n<p>Thomas</p>",873 "post_number": 4,874 "post_type": 1,875 "posts_count": 8,876 "updated_at": "2020-07-15T11:42:28.471Z",877 "reply_count": 1,878 "reply_to_post_number": 3,879 "quote_count": 0,880 "incoming_link_count": 19,881 "reads": 110,882 "readers_count": 109,883 "score": 167.0,884 "yours": false,885 "topic_id": 89290,886 "topic_slug": "how-to-modify-gradients-in-place",887 "display_username": "Thomas V",888 "primary_group_name": null,889 "flair_name": null,890 "flair_url": null,891 "flair_bg_color": null,892 "flair_color": null,893 "flair_group_id": null,894 "badges_granted": [],895 "version": 1,896 "can_edit": false,897 "can_delete": false,898 "can_recover": false,899 "can_see_hidden_post": false,900 "can_wiki": false,901 "read": true,902 "user_title": null,903 "reply_to_user": {904 "id": 18090,905 "username": "Florian_Dietz",906 "name": "Florian Dietz",907 "avatar_template": "/user_avatar/discuss.pytorch.org/florian_dietz/{size}/11502_2.png"908 },909 "bookmarked": false,910 "actions_summary": [911 {912 "id": 2,913 "count": 1914 }915 ],916 "moderator": false,917 "admin": false,918 "staff": false,919 "user_id": 616,920 "hidden": false,921 "trust_level": 2,922 "deleted_at": null,923 "user_deleted": false,924 "edit_reason": null,925 "can_view_edit_history": true,926 "wiki": false,927 "post_url": "/t/how-to-modify-gradients-in-place/89290/4",928 "can_accept_answer": false,929 "can_unaccept_answer": false,930 "accepted_answer": false,931 "topic_accepted_answer": null932 },933 {934 "id": 212580,935 "name": "Florian Dietz",936 "username": "Florian_Dietz",937 "avatar_template": "/user_avatar/discuss.pytorch.org/florian_dietz/{size}/11502_2.png",938 "created_at": "2020-07-15T12:28:02.336Z",939 "cooked": "<p>That’s interesting. I didn’t know there was a difference between <code>torch.autograd.grad</code> and <code>torch.autograd.backward</code>. It looks like the first one is functional, while the second has side-effects. Is that correct? Is there anything you would recommend to read to understand this better? I’m worried I might break something if I make a switch at this point without fully understanding everything.</p>",940 "post_number": 5,941 "post_type": 1,942 "posts_count": 8,943 "updated_at": "2020-07-15T12:28:02.336Z",944 "reply_count": 1,945 "reply_to_post_number": 4,946 "quote_count": 0,947 "incoming_link_count": 4,948 "reads": 96,949 "readers_count": 95,950 "score": 44.2,951 "yours": false,952 "topic_id": 89290,953 "topic_slug": "how-to-modify-gradients-in-place",954 "display_username": "Florian Dietz",955 "primary_group_name": null,956 "flair_name": null,957 "flair_url": null,958 "flair_bg_color": null,959 "flair_color": null,960 "flair_group_id": null,961 "badges_granted": [],962 "version": 1,963 "can_edit": false,964 "can_delete": false,965 "can_recover": false,966 "can_see_hidden_post": false,967 "can_wiki": false,968 "read": true,969 "user_title": null,970 "reply_to_user": {971 "id": 616,972 "username": "tom",973 "name": "Thomas V",974 "avatar_template": "/user_avatar/discuss.pytorch.org/tom/{size}/3162_2.png"975 },976 "bookmarked": false,977 "actions_summary": [],978 "moderator": false,979 "admin": false,980 "staff": false,981 "user_id": 18090,982 "hidden": false,983 "trust_level": 1,984 "deleted_at": null,985 "user_deleted": false,986 "edit_reason": null,987 "can_view_edit_history": true,988 "wiki": false,989 "post_url": "/t/how-to-modify-gradients-in-place/89290/5",990 "can_accept_answer": false,991 "can_unaccept_answer": false,992 "accepted_answer": false,993 "topic_accepted_answer": null994 },995 {996 "id": 212595,997 "name": "Thomas V",998 "username": "tom",999 "avatar_template": "/user_avatar/discuss.pytorch.org/tom/{size}/3162_2.png",1000 "created_at": "2020-07-15T13:47:31.171Z",1001 "cooked": "<p>Yeah, so basically, the difference is “only” functional vs. side-effects.<br>\nFor optimizers, you need to have <code>.backward</code> as .grad is what they consume.<br>\nBut for other purposes it would seem that <code>autograd.grad</code> makes is clear that the result is yours.<br>\nProbably a very careful reading of the documentation (the Autograd Mechanics note and torch.autograd come to mind).</p>\n<p>Best regards</p>\n<p>Thomas</p>",1002 "post_number": 6,1003 "post_type": 1,1004 "posts_count": 8,1005 "updated_at": "2020-07-15T13:47:31.171Z",1006 "reply_count": 1,1007 "reply_to_post_number": 5,1008 "quote_count": 0,1009 "incoming_link_count": 13,1010 "reads": 94,1011 "readers_count": 93,1012 "score": 103.8,1013 "yours": false,1014 "topic_id": 89290,1015 "topic_slug": "how-to-modify-gradients-in-place",1016 "display_username": "Thomas V",1017 "primary_group_name": null,1018 "flair_name": null,1019 "flair_url": null,1020 "flair_bg_color": null,1021 "flair_color": null,1022 "flair_group_id": null,1023 "badges_granted": [],1024 "version": 1,1025 "can_edit": false,1026 "can_delete": false,1027 "can_recover": false,1028 "can_see_hidden_post": false,1029 "can_wiki": false,1030 "read": true,1031 "user_title": null,1032 "reply_to_user": {1033 "id": 18090,1034 "username": "Florian_Dietz",1035 "name": "Florian Dietz",1036 "avatar_template": "/user_avatar/discuss.pytorch.org/florian_dietz/{size}/11502_2.png"1037 },1038 "bookmarked": false,1039 "actions_summary": [1040 {1041 "id": 2,1042 "count": 11043 }1044 ],1045 "moderator": false,1046 "admin": false,1047 "staff": false,1048 "user_id": 616,1049 "hidden": false,1050 "trust_level": 2,1051 "deleted_at": null,1052 "user_deleted": false,1053 "edit_reason": null,1054 "can_view_edit_history": true,1055 "wiki": false,1056 "post_url": "/t/how-to-modify-gradients-in-place/89290/6",1057 "can_accept_answer": false,1058 "can_unaccept_answer": false,1059 "accepted_answer": false,1060 "topic_accepted_answer": null1061 },1062 {1063 "id": 212609,1064 "name": "Florian Dietz",1065 "username": "Florian_Dietz",1066 "avatar_template": "/user_avatar/discuss.pytorch.org/florian_dietz/{size}/11502_2.png",1067 "created_at": "2020-07-15T14:43:55.102Z",1068 "cooked": "<p>Thanks! That’s good to know. Now I just need to figure out if that’s worth all the refactoring…</p>",1069 "post_number": 7,1070 "post_type": 1,1071 "posts_count": 8,1072 "updated_at": "2020-07-15T14:43:55.102Z",1073 "reply_count": 1,1074 "reply_to_post_number": 6,1075 "quote_count": 0,1076 "incoming_link_count": 2,1077 "reads": 88,1078 "readers_count": 87,1079 "score": 27.6,1080 "yours": false,1081 "topic_id": 89290,1082 "topic_slug": "how-to-modify-gradients-in-place",1083 "display_username": "Florian Dietz",1084 "primary_group_name": null,1085 "flair_name": null,1086 "flair_url": null,1087 "flair_bg_color": null,1088 "flair_color": null,1089 "flair_group_id": null,1090 "badges_granted": [],1091 "version": 1,1092 "can_edit": false,1093 "can_delete": false,1094 "can_recover": false,1095 "can_see_hidden_post": false,1096 "can_wiki": false,1097 "read": true,1098 "user_title": null,1099 "reply_to_user": {1100 "id": 616,1101 "username": "tom",1102 "name": "Thomas V",1103 "avatar_template": "/user_avatar/discuss.pytorch.org/tom/{size}/3162_2.png"1104 },1105 "bookmarked": false,1106 "actions_summary": [],1107 "moderator": false,1108 "admin": false,1109 "staff": false,1110 "user_id": 18090,1111 "hidden": false,1112 "trust_level": 1,1113 "deleted_at": null,1114 "user_deleted": false,1115 "edit_reason": null,1116 "can_view_edit_history": true,1117 "wiki": false,1118 "post_url": "/t/how-to-modify-gradients-in-place/89290/7",1119 "can_accept_answer": false,1120 "can_unaccept_answer": false,1121 "accepted_answer": false,1122 "topic_accepted_answer": null1123 },1124 {1125 "id": 216263,1126 "name": "Florian Dietz",1127 "username": "Florian_Dietz",1128 "avatar_template": "/user_avatar/discuss.pytorch.org/florian_dietz/{size}/11502_2.png",1129 "created_at": "2020-07-28T08:41:13.721Z",1130 "cooked": "<p>Quick update: In the meantime, I have noticed that my code contained a bug. I tried to apply a loss to a calculation after some parameters used in it had already been modified by another calculation. This is what caused the error message. The question I asked is still valid, and the answers are still useful. But just in case somebody wonders, the reason for the error message was not the in-place modification of gradients.</p>",1131 "post_number": 8,1132 "post_type": 1,1133 "posts_count": 8,1134 "updated_at": "2020-07-28T08:41:13.721Z",1135 "reply_count": 0,1136 "reply_to_post_number": 7,1137 "quote_count": 0,1138 "incoming_link_count": 8,1139 "reads": 84,1140 "readers_count": 83,1141 "score": 56.8,1142 "yours": false,1143 "topic_id": 89290,1144 "topic_slug": "how-to-modify-gradients-in-place",1145 "display_username": "Florian Dietz",1146 "primary_group_name": null,1147 "flair_name": null,1148 "flair_url": null,1149 "flair_bg_color": null,1150 "flair_color": null,1151 "flair_group_id": null,1152 "badges_granted": [],1153 "version": 1,1154 "can_edit": false,1155 "can_delete": false,1156 "can_recover": false,1157 "can_see_hidden_post": false,1158 "can_wiki": false,1159 "read": true,1160 "user_title": null,1161 "reply_to_user": {1162 "id": 18090,1163 "username": "Florian_Dietz",1164 "name": "Florian Dietz",1165 "avatar_template": "/user_avatar/discuss.pytorch.org/florian_dietz/{size}/11502_2.png"1166 },1167 "bookmarked": false,1168 "actions_summary": [],1169 "moderator": false,1170 "admin": false,1171 "staff": false,1172 "user_id": 18090,1173 "hidden": false,1174 "trust_level": 1,1175 "deleted_at": null,1176 "user_deleted": false,1177 "edit_reason": null,1178 "can_view_edit_history": true,1179 "wiki": false,1180 "post_url": "/t/how-to-modify-gradients-in-place/89290/8",1181 "can_accept_answer": false,1182 "can_unaccept_answer": false,1183 "accepted_answer": false,1184 "topic_accepted_answer": null1185 }1186 ],1187 "stream": [1188 212502,1189 212543,1190 212561,1191 212566,1192 212580,1193 212595,1194 212609,1195 2162631196 ]1197 },1198 "timeline_lookup": [1199 [1200 1,