🕷️ Crawler Inspector

URL Lookup

Direct Parameter Lookup

Raw Queries and Responses

1. Shard Calculation

Query:

Response:

Calculated Shard: 169 (from laksa104)

2. Crawled Status Check

Query:

curl -X POST \
  'http://laksa169.int.ahrefs:8124/' \
  -H 'Content-Type: text/plain' \
  -H 'X-ClickHouse-Database: crawler3' \
  -H 'Authorization: Basic YXBpOg==' \
  -d 'SELECT getAhrefsURLFromUnparsed(src_unparsed) AS found_url, ifNull(toUnixTimestamp(download_stamp), 0) AS crawl_time, ifNull(toUnixTimestamp(props_url_first_seen), 0) AS first_indexed_time, download_http_code AS http_code, src_unparsed AS src_unparsed, src_root_hash AS src_root_hash, history_drop_reason AS history_drop_reason, meta_title AS meta_title, meta_descriptions AS meta_descriptions, attrs_boilerpipe_text AS attrs_boilerpipe_text, attrs_markdown AS attrs_markdown, attrs_readable_markdown AS attrs_readable_markdown, meta_canonical AS meta_canonical FROM crawler3.page_info_local FINAL PREWHERE (src_root_hash, src_unparsed) IN ((getAhrefsRootHashFromUnparsed(getAhrefsUnparsedNoserviceFromURL(\'https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu\')), getAhrefsUnparsedNoserviceFromURL(\'https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu\'))) FORMAT JSONEachRow'

Response:

{"found_url":"https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu","crawl_time":1775169942,"first_indexed_time":1630759010,"http_code":200,"src_unparsed":"com,stackoverflow!\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu s443","src_root_hash":"714406497480128969","history_drop_reason":null,"meta_title":"python - How does one use Pytorch (+ cuda) with an A100 GPU? - Stack Overflow","meta_descriptions":[],"attrs_boilerpipe_text":"I was trying to use my current code with an A100 gpu but I get this error:\n---> backend=\n'nccl'\n\/home\/miranda9\/miniconda3\/envs\/metalearningpy1\n.7\n.1\nc10\n.2\n\/lib\/python3\n.8\n\/site-packages\/torch\/cuda\/__init__.py:\n104\n: UserWarning: \nA100-SXM4-40GB\nwith\nCUDA capability sm_80\nis\nnot\ncompatible\nwith\nthe current PyTorch installation.\nThe current PyTorch install supports CUDA capabilities sm_37 sm_50 sm_60 sm_61 sm_70 sm_75 compute_37.\nIf you want to use the A100-SXM4-40GB GPU\nwith\nPyTorch, please check the instructions at https:\/\/pytorch.org\/get-started\/locally\/\nwhich is reather confusing because it points to the usual pytorch installation but doesn't tell me which combination of pytorch version + cuda version to use for my specific hardware (A100). What is the right way to install pytorch for an A100?\nThese are some versions I've tried:\n# conda install -y pytorch==1.8.0 torchvision cudatoolkit=10.2 -c pytorch\n# conda install -y pytorch torchvision cudatoolkit=10.2 -c pytorch\n#conda install -y pytorch==1.7.1 torchvision torchaudio cudatoolkit=10.2 -c pytorch -c conda-forge\n# conda install -y pytorch==1.6.0 torchvision cudatoolkit=10.2 -c pytorch\n#conda install -y pytorch==1.7.1 torchvision torchaudio cudatoolkit=11.1 -c pytorch -c conda-forge\n# conda install pytorch torchvision torchaudio cudatoolkit=11.0 -c pytorch\n# conda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c conda-forge\n# conda install -y pytorch torchvision cudatoolkit=9.2 -c pytorch # For Nano, CC\n# conda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c conda-forge\nnote that this can be subtle because I've had this error with this machine + pytorch version in the past:\nHow to solve the famous `unhandled cuda error, NCCL version 2.7.8` error?\nBonus 1:\nI still have errors:\nncclSystemError: System call (socket, malloc, munmap, etc) failed.\nTraceback (most recent call last):\n  File\n\"\/home\/miranda9\/diversity-for-predictive-success-of-meta-learning\/div_src\/diversity_src\/experiment_mains\/main_dist_maml_l2l.py\"\n, line\n1423\n,\nin\n<module>\n    main()\n  File\n\"\/home\/miranda9\/diversity-for-predictive-success-of-meta-learning\/div_src\/diversity_src\/experiment_mains\/main_dist_maml_l2l.py\"\n, line\n1365\n,\nin\nmain\n    train(args=args)\n  File\n\"\/home\/miranda9\/diversity-for-predictive-success-of-meta-learning\/div_src\/diversity_src\/experiment_mains\/main_dist_maml_l2l.py\"\n, line\n1385\n,\nin\ntrain\n    args.opt = move_opt_to_cherry_opt_and_sync_params(args)\nif\nis_running_parallel(args.rank)\nelse\nargs.opt\n  File\n\"\/home\/miranda9\/ultimate-utils\/ultimate-utils-proj-src\/uutils\/torch_uu\/distributed.py\"\n, line\n456\n,\nin\nmove_opt_to_cherry_opt_and_sync_params\n    args.opt = cherry.optim.Distributed(args.model.parameters(), opt=args.opt, sync=syn)\n  File\n\"\/home\/miranda9\/miniconda3\/envs\/meta_learning_a100\/lib\/python3.9\/site-packages\/cherry\/optim.py\"\n, line\n62\n,\nin\n__init__\n    self.sync_parameters()\n  File\n\"\/home\/miranda9\/miniconda3\/envs\/meta_learning_a100\/lib\/python3.9\/site-packages\/cherry\/optim.py\"\n, line\n78\n,\nin\nsync_parameters\n    dist.broadcast(p.data, src=root)\n  File\n\"\/home\/miranda9\/miniconda3\/envs\/meta_learning_a100\/lib\/python3.9\/site-packages\/torch\/distributed\/distributed_c10d.py\"\n, line\n1090\n,\nin\nbroadcast\n    work = default_pg.broadcast([tensor], opts)\nRuntimeError: NCCL error\nin\n: ..\/torch\/lib\/c10d\/ProcessGroupNCCL.cpp:\n911\n, unhandled system error, NCCL version\n2.7\n.8\none of the answers suggested to have nvcca & pytorch.version.cuda to match but they do not:\n(meta_learning_a100) [miranda9@hal-dgx ~]$ python -c\n\"import torch;print(torch.version.cuda)\"\n11.1\n(meta_learning_a100) [miranda9@hal-dgx ~]$ nvcc -V\nnvcc: NVIDIA (R) Cuda compiler driver\nCopyright (c)\n2005\n-\n2020\nNVIDIA Corporation\nBuilt on Wed_Jul_22_19:09:09_PDT_2020\nCuda compilation tools, release\n11.0\n, V11\n.0\n.221\nBuild cuda_11\n.0\n_bu.TC445_37\n.28845127_0\nHow do I match them? I this the error? Can someone display their pip, conda and nvcca version to see what set up works?\nMore error messages:\nhal-dgx:\n21797\n:\n21797\n[\n0\n] NCCL INFO Bootstrap : Using [\n0\n]enp226s0:\n141.142\n.153\n.83\n<\n0\n> [\n1\n]virbr0:\n192.168\n.122\n.1\n<\n0\n>\nhal-dgx:\n21797\n:\n21797\n[\n0\n] NCCL INFO NET\/Plugin : No plugin found (libnccl-net.so), using internal implementation\nhal-dgx:\n21797\n:\n21797\n[\n0\n] NCCL INFO NET\/IB : Using [\n0\n]mlx5_0:\n1\n\/IB [\n1\n]mlx5_1:\n1\n\/IB [\n2\n]mlx5_2:\n1\n\/IB [\n3\n]mlx5_3:\n1\n\/IB [\n4\n]mlx5_4:\n1\n\/IB [\n5\n]mlx5_5:\n1\n\/IB [\n6\n]mlx5_6:\n1\n\/IB [\n7\n]mlx5_7:\n1\n\/IB ; OOB enp226s0:\n141.142\n.153\n.83\n<\n0\n>\nhal-dgx:\n21797\n:\n21797\n[\n0\n] NCCL INFO Using network IB\nNCCL version\n2.7\n.8\n+cuda11\n.1\nhal-dgx:\n21805\n:\n21805\n[\n2\n] NCCL INFO Bootstrap : Using [\n0\n]enp226s0:\n141.142\n.153\n.83\n<\n0\n> [\n1\n]virbr0:\n192.168\n.122\n.1\n<\n0\n>\nhal-dgx:\n21799\n:\n21799\n[\n1\n] NCCL INFO Bootstrap : Using [\n0\n]enp226s0:\n141.142\n.153\n.83\n<\n0\n> [\n1\n]virbr0:\n192.168\n.122\n.1\n<\n0\n>\nhal-dgx:\n21805\n:\n21805\n[\n2\n] NCCL INFO NET\/Plugin : No plugin found (libnccl-net.so), using internal implementation\nhal-dgx:\n21799\n:\n21799\n[\n1\n] NCCL INFO NET\/Plugin : No plugin found (libnccl-net.so), using internal implementation\nhal-dgx:\n21811\n:\n21811\n[\n3\n] NCCL INFO Bootstrap : Using [\n0\n]enp226s0:\n141.142\n.153\n.83\n<\n0\n> [\n1\n]virbr0:\n192.168\n.122\n.1\n<\n0\n>\nhal-dgx:\n21811\n:\n21811\n[\n3\n] NCCL INFO NET\/Plugin : No plugin found (libnccl-net.so), using internal implementation\nhal-dgx:\n21811\n:\n21811\n[\n3\n] NCCL INFO NET\/IB : Using [\n0\n]mlx5_0:\n1\n\/IB [\n1\n]mlx5_1:\n1\n\/IB [\n2\n]mlx5_2:\n1\n\/IB [\n3\n]mlx5_3:\n1\n\/IB [\n4\n]mlx5_4:\n1\n\/IB [\n5\n]mlx5_5:\n1\n\/IB [\n6\n]mlx5_6:\n1\n\/IB [\n7\n]mlx5_7:\n1\n\/IB ; OOB enp226s0:\n141.142\n.153\n.83\n<\n0\n>\nhal-dgx:\n21811\n:\n21811\n[\n3\n] NCCL INFO Using network IB\nhal-dgx:\n21799\n:\n21799\n[\n1\n] NCCL INFO NET\/IB : Using [\n0\n]mlx5_0:\n1\n\/IB [\n1\n]mlx5_1:\n1\n\/IB [\n2\n]mlx5_2:\n1\n\/IB [\n3\n]mlx5_3:\n1\n\/IB [\n4\n]mlx5_4:\n1\n\/IB [\n5\n]mlx5_5:\n1\n\/IB [\n6\n]mlx5_6:\n1\n\/IB [\n7\n]mlx5_7:\n1\n\/IB ; OOB enp226s0:\n141.142\n.153\n.83\n<\n0\n>\nhal-dgx:\n21805\n:\n21805\n[\n2\n] NCCL INFO NET\/IB : Using [\n0\n]mlx5_0:\n1\n\/IB [\n1\n]mlx5_1:\n1\n\/IB [\n2\n]mlx5_2:\n1\n\/IB [\n3\n]mlx5_3:\n1\n\/IB [\n4\n]mlx5_4:\n1\n\/IB [\n5\n]mlx5_5:\n1\n\/IB [\n6\n]mlx5_6:\n1\n\/IB [\n7\n]mlx5_7:\n1\n\/IB ; OOB enp226s0:\n141.142\n.153\n.83\n<\n0\n>\nhal-dgx:\n21799\n:\n21799\n[\n1\n] NCCL INFO Using network IB\nhal-dgx:\n21805\n:\n21805\n[\n2\n] NCCL INFO Using network IB\n\nhal-dgx:\n21797\n:\n27906\n[\n0\n] misc\/ibvwrap.cc:\n280\nNCCL WARN Call to ibv_create_qp failed\nhal-dgx:\n21797\n:\n27906\n[\n0\n] NCCL INFO transport\/net_ib.cc:\n360\n->\n2\nhal-dgx:\n21797\n:\n27906\n[\n0\n] NCCL INFO transport\/net_ib.cc:\n437\n->\n2\nhal-dgx:\n21797\n:\n27906\n[\n0\n] NCCL INFO include\/net.h:\n21\n->\n2\nhal-dgx:\n21797\n:\n27906\n[\n0\n] NCCL INFO include\/net.h:\n51\n->\n2\nhal-dgx:\n21797\n:\n27906\n[\n0\n] NCCL INFO init.cc:\n300\n->\n2\nhal-dgx:\n21797\n:\n27906\n[\n0\n] NCCL INFO init.cc:\n566\n->\n2\nhal-dgx:\n21797\n:\n27906\n[\n0\n] NCCL INFO init.cc:\n840\n->\n2\nhal-dgx:\n21797\n:\n27906\n[\n0\n] NCCL INFO group.cc:\n73\n->\n2\n[Async thread]\n\nhal-dgx:\n21811\n:\n27929\n[\n3\n] misc\/ibvwrap.cc:\n280\nNCCL WARN Call to ibv_create_qp failed\nhal-dgx:\n21811\n:\n27929\n[\n3\n] NCCL INFO transport\/net_ib.cc:\n360\n->\n2\nhal-dgx:\n21811\n:\n27929\n[\n3\n] NCCL INFO transport\/net_ib.cc:\n437\n->\n2\nhal-dgx:\n21811\n:\n27929\n[\n3\n] NCCL INFO include\/net.h:\n21\n->\n2\nhal-dgx:\n21811\n:\n27929\n[\n3\n] NCCL INFO include\/net.h:\n51\n->\n2\nhal-dgx:\n21811\n:\n27929\n[\n3\n] NCCL INFO init.cc:\n300\n->\n2\nhal-dgx:\n21811\n:\n27929\n[\n3\n] NCCL INFO init.cc:\n566\n->\n2\nhal-dgx:\n21811\n:\n27929\n[\n3\n] NCCL INFO init.cc:\n840\n->\n2\nhal-dgx:\n21811\n:\n27929\n[\n3\n] NCCL INFO group.cc:\n73\n->\n2\n[Async thread]\nafter putting\nimport\nos\nos.environ[\n\"NCCL_DEBUG\"\n] =\n\"INFO\"","attrs_markdown":"# ![site logo](https:\/\/stackoverflow.com\/Content\/Img\/SE-logo75.png)\nBy clicking “Sign up”, you agree to our [terms of service](https:\/\/stackoverflow.com\/legal\/terms-of-service\/public) and acknowledge you have read our [privacy policy](https:\/\/stackoverflow.com\/legal\/privacy-policy).\n\n# OR\n\nAlready have an account? [Log in](https:\/\/stackoverflow.com\/users\/login)\n\n[Skip to main content](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#content)\n\n[Stack Overflow](https:\/\/stackoverflow.com\/)\n\n1. [About](https:\/\/stackoverflow.co\/)\n2. Products\n3. [For Teams](https:\/\/stackoverflow.co\/internal\/)\n\n1. [Stack Internal Implement a knowledge platform layer to power your enterprise and AI tools.](https:\/\/stackoverflow.co\/internal\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=stack-overflow-for-teams)\n2. [Stack Data Licensing Get access to top-class technical expertise with trusted & attributed content.](https:\/\/stackoverflow.co\/data-licensing\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=overflow-api)\n3. [Stack Ads Connect your brand to the world’s most trusted technologist communities.](https:\/\/stackoverflow.co\/advertising\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=stack-overflow-advertising)\n4. [Releases Keep up-to-date on features we add to Stack Overflow and Stack Internal.](https:\/\/stackoverflow.blog\/releases\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=releases)\n5. [About the company](https:\/\/stackoverflow.co\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=about-the-company) [Visit the blog](https:\/\/stackoverflow.blog\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=blog)\n\n1. ### [current community](https:\/\/stackoverflow.com\/)\n   - [Stack Overflow](https:\/\/stackoverflow.com\/)\n     [help](https:\/\/stackoverflow.com\/help) [chat](https:\/\/chat.stackoverflow.com\/?tab=explore)\n   - [Meta Stack Overflow](https:\/\/meta.stackoverflow.com\/)\n   ### your communities\n   [Sign up](https:\/\/stackoverflow.com\/users\/signup?ssrc=site_switcher&returnurl=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F66992585%2Fhow-does-one-use-pytorch-cuda-with-an-a100-gpu) or [log in](https:\/\/stackoverflow.com\/users\/login?ssrc=site_switcher&returnurl=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F66992585%2Fhow-does-one-use-pytorch-cuda-with-an-a100-gpu) to customize your list.\n   \n   ### [more stack exchange communities](https:\/\/stackexchange.com\/sites)\n   [company blog](https:\/\/stackoverflow.blog\/)\n2. [Log in](https:\/\/stackoverflow.com\/users\/login?ssrc=head&returnurl=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F66992585%2Fhow-does-one-use-pytorch-cuda-with-an-a100-gpu)\n3. [Sign up](https:\/\/stackoverflow.com\/users\/signup?ssrc=head&returnurl=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F66992585%2Fhow-does-one-use-pytorch-cuda-with-an-a100-gpu)\n\n# Let's set up your homepage Select a few topics you're interested in:\npython\n\njavascript\n\nc\\#\n\nreactjs\n\njava\n\nandroid\n\nhtml\n\nflutter\n\nc++\n\nnode.js\n\ntypescript\n\ncss\n\nr\n\nphp\n\nangular\n\nnext.js\n\nspring-boot\n\nmachine-learning\n\nsql\n\nexcel\n\nios\n\nazure\n\ndocker\n\nOr search from our full list:\n\n- javascript\n- python\n- java\n- c\\#\n- php\n- android\n- html\n- jquery\n- c++\n- css\n- ios\n- sql\n- mysql\n- r\n- reactjs\n- node.js\n- arrays\n- c\n- asp.net\n- json\n- python-3.x\n- .net\n- ruby-on-rails\n- sql-server\n- swift\n- django\n- angular\n- objective-c\n- excel\n- pandas\n- angularjs\n- regex\n- typescript\n- ruby\n- linux\n- ajax\n- iphone\n- vba\n- xml\n- laravel\n- spring\n- asp.net-mvc\n- database\n- wordpress\n- string\n- flutter\n- postgresql\n- mongodb\n- wpf\n- windows\n- xcode\n- amazon-web-services\n- bash\n- git\n- oracle-database\n- spring-boot\n- dataframe\n- azure\n- firebase\n- list\n- multithreading\n- docker\n- vb.net\n- react-native\n- eclipse\n- algorithm\n- powershell\n- macos\n- visual-studio\n- numpy\n- image\n- forms\n- scala\n- function\n- vue.js\n- performance\n- twitter-bootstrap\n- selenium\n- winforms\n- kotlin\n- loops\n- dart\n- express\n- hibernate\n- sqlite\n- matlab\n- python-2.7\n- shell\n- rest\n- apache\n- entity-framework\n- android-studio\n- csv\n- maven\n- linq\n- qt\n- dictionary\n- unit-testing\n- asp.net-core\n- facebook\n- apache-spark\n- tensorflow\n- file\n- swing\n- class\n- unity-game-engine\n- sorting\n- date\n- authentication\n- go\n- symfony\n- t-sql\n- opencv\n- matplotlib\n- .htaccess\n- google-chrome\n- for-loop\n- datetime\n- codeigniter\n- perl\n- http\n- validation\n- sockets\n- google-maps\n- object\n- uitableview\n- xaml\n- oop\n- visual-studio-code\n- if-statement\n- cordova\n- ubuntu\n- web-services\n- email\n- android-layout\n- github\n- spring-mvc\n- elasticsearch\n- kubernetes\n- selenium-webdriver\n- ms-access\n- ggplot2\n- user-interface\n- parsing\n- pointers\n- google-sheets\n- c++11\n- security\n- machine-learning\n- google-apps-script\n- ruby-on-rails-3\n- templates\n- flask\n- nginx\n- variables\n- exception\n- sql-server-2008\n- gradle\n- debugging\n- tkinter\n- delphi\n- listview\n- jpa\n- asynchronous\n- haskell\n- web-scraping\n- pdf\n- jsp\n- ssl\n- amazon-s3\n- google-cloud-platform\n- testing\n- jenkins\n- xamarin\n- wcf\n- batch-file\n- generics\n- npm\n- ionic-framework\n- network-programming\n- unix\n- recursion\n- google-app-engine\n- mongoose\n- visual-studio-2010\n- .net-core\n- android-fragments\n- assembly\n- animation\n- math\n- svg\n- rust\n- session\n- intellij-idea\n- hadoop\n- join\n- curl\n- winapi\n- django-models\n- laravel-5\n- next.js\n- url\n- heroku\n- http-redirect\n- tomcat\n- inheritance\n- google-cloud-firestore\n- webpack\n- gcc\n- image-processing\n- swiftui\n- keras\n- asp.net-mvc-4\n- logging\n- dom\n- matrix\n- pyspark\n- actionscript-3\n- button\n- post\n- optimization\n- firebase-realtime-database\n- jquery-ui\n- cocoa\n- xpath\n- iis\n- web\n- d3.js\n- javafx\n- firefox\n- xslt\n- internet-explorer\n- caching\n- select\n- asp.net-mvc-3\n- opengl\n- events\n- asp.net-web-api\n- plot\n- dplyr\n- encryption\n- magento\n- stored-procedures\n- search\n- amazon-ec2\n- ruby-on-rails-4\n- memory\n- canvas\n- multidimensional-array\n- audio\n- random\n- jsf\n- vector\n- redux\n- cookies\n- input\n- facebook-graph-api\n- flash\n- indexing\n- xamarin.forms\n- arraylist\n- ipad\n- cocoa-touch\n- data-structures\n- video\n- model-view-controller\n- azure-devops\n- serialization\n- apache-kafka\n- jdbc\n- razor\n- woocommerce\n- awk\n- routes\n- mod-rewrite\n- servlets\n- excel-formula\n- beautifulsoup\n- filter\n- iframe\n- docker-compose\n- design-patterns\n- aws-lambda\n- text\n- visual-c++\n- django-rest-framework\n- cakephp\n- mobile\n- android-intent\n- struct\n- react-hooks\n- methods\n- groovy\n- mvvm\n- lambda\n- ssh\n- checkbox\n- time\n- ecmascript-6\n- grails\n- google-chrome-extension\n- installation\n- cmake\n- sharepoint\n- shiny\n- spring-security\n- jakarta-ee\n- plsql\n- android-recyclerview\n- core-data\n- types\n- sed\n- meteor\n- android-activity\n- bootstrap-4\n- activerecord\n- websocket\n- replace\n- graph\n- group-by\n- scikit-learn\n- vim\n- file-upload\n- boost\n- junit\n- memory-management\n- sass\n- async-await\n- import\n- deep-learning\n- error-handling\n- eloquent\n- dynamic\n- dependency-injection\n- soap\n- silverlight\n- layout\n- apache-spark-sql\n- charts\n- deployment\n- browser\n- gridview\n- svn\n- while-loop\n- google-bigquery\n- vuejs2\n- highcharts\n- dll\n- ffmpeg\n- view\n- foreach\n- makefile\n- plugins\n- redis\n- c\\#-4.0\n- reporting-services\n- jupyter-notebook\n- unicode\n- merge\n- reflection\n- https\n- server\n- google-maps-api-3\n- twitter\n- oauth-2.0\n- extjs\n- terminal\n- axios\n- pip\n- split\n- cmd\n- pytorch\n- encoding\n- django-views\n- collections\n- database-design\n- hash\n- netbeans\n- automation\n- data-binding\n- ember.js\n- build\n- tcp\n- pdo\n- sqlalchemy\n- apache-flex\n- entity-framework-core\n- concurrency\n- command-line\n- spring-data-jpa\n- printing\n- react-redux\n- java-8\n- lua\n- html-table\n- neo4j\n- ansible\n- service\n- jestjs\n- mysqli\n- enums\n- parameters\n- flexbox\n- material-ui\n- module\n- promise\n- visual-studio-2012\n- outlook\n- web-applications\n- firebase-authentication\n- uwp\n- webview\n- jquery-mobile\n- utf-8\n- datatable\n- python-requests\n- parallel-processing\n- colors\n- drop-down-menu\n- scipy\n- scroll\n- tfs\n- hive\n- count\n- syntax\n- ms-word\n- twitter-bootstrap-3\n- ssis\n- fonts\n- rxjs\n- constructor\n- google-analytics\n- file-io\n- paypal\n- three.js\n- powerbi\n- graphql\n- cassandra\n- discord\n- graphics\n- compiler-errors\n- gwt\n- react-router\n- socket.io\n- backbone.js\n- solr\n- memory-leaks\n- url-rewriting\n- datatables\n- nlp\n- terraform\n- oauth\n- datagridview\n- drupal\n- zend-framework\n- oracle11g\n- knockout.js\n- triggers\n- interface\n- neural-network\n- django-forms\n- casting\n- angular-material\n- jmeter\n- linked-list\n- google-api\n- path\n- timer\n- arduino\n- django-templates\n- orm\n- proxy\n- directory\n- windows-phone-7\n- parse-platform\n- visual-studio-2015\n- cron\n- conditional-statements\n- push-notification\n- functional-programming\n- primefaces\n- pagination\n- model\n- jar\n- xamarin.android\n- hyperlink\n- uiview\n- vbscript\n- visual-studio-2013\n- google-cloud-functions\n- azure-active-directory\n- gitlab\n- jwt\n- download\n- swift3\n- sql-server-2005\n- process\n- configuration\n- rspec\n- pygame\n- properties\n- callback\n- combobox\n- windows-phone-8\n- linux-kernel\n- safari\n- scrapy\n- permissions\n- emacs\n- x86\n- clojure\n- scripting\n- raspberry-pi\n- scope\n- io\n- azure-functions\n- compilation\n- expo\n- responsive-design\n- mongodb-query\n- nhibernate\n- angularjs-directive\n- reference\n- request\n- bluetooth\n- binding\n- architecture\n- dns\n- playframework\n- 3d\n- pyqt\n- version-control\n- discord.js\n- doctrine-orm\n- package\n- f\\#\n- rubygems\n- get\n- sql-server-2012\n- autocomplete\n- tree\n- openssl\n- datepicker\n- kendo-ui\n- jackson\n- yii\n- controller\n- grep\n- nested\n- xamarin.ios\n- static\n- null\n- transactions\n- statistics\n- datagrid\n- active-directory\n- dockerfile\n- uiviewcontroller\n- webforms\n- sas\n- discord.py\n- computer-vision\n- phpmyadmin\n- notifications\n- duplicates\n- pycharm\n- mocking\n- youtube\n- yaml\n- nullpointerexception\n- menu\n- blazor\n- sum\n- plotly\n- bitmap\n- visual-studio-2008\n- asp.net-mvc-5\n- floating-point\n- css-selectors\n- yii2\n- stl\n- electron\n- android-listview\n- jsf-2\n- time-series\n- cryptography\n- ant\n- character-encoding\n- hashmap\n- stream\n- msbuild\n- asp.net-core-mvc\n- sdk\n- google-drive-api\n- jboss\n- selenium-chromedriver\n- joomla\n- devise\n- navigation\n- cors\n- cuda\n- anaconda\n- frontend\n- background\n- multiprocessing\n- binary\n- pyqt5\n- camera\n- iterator\n- linq-to-sql\n- mariadb\n- onclick\n- android-jetpack-compose\n- ios7\n- microsoft-graph-api\n- android-asynctask\n- rabbitmq\n- tabs\n- amazon-dynamodb\n- environment-variables\n- laravel-4\n- uicollectionview\n- insert\n- linker\n- xsd\n- coldfusion\n- console\n- continuous-integration\n- upload\n- textview\n- ftp\n- opengl-es\n- operating-system\n- macros\n- mockito\n- localization\n- formatting\n- xml-parsing\n- json.net\n- type-conversion\n- vuejs3\n- data.table\n- kivy\n- timestamp\n- integer\n- calendar\n- segmentation-fault\n- android-ndk\n- prolog\n- char\n- drag-and-drop\n- crash\n- jasmine\n- automated-tests\n- dependencies\n- geometry\n- azure-pipelines\n- fortran\n- android-gradle-plugin\n- itext\n- sprite-kit\n- mfc\n- header\n- attributes\n- nosql\n- firebase-cloud-messaging\n- format\n- nuxt.js\n- odoo\n- db2\n- jquery-plugins\n- event-handling\n- julia\n- jenkins-pipeline\n- leaflet\n- annotations\n- flutter-layout\n- nestjs\n- keyboard\n- postman\n- arm\n- textbox\n- stripe-payments\n- visual-studio-2017\n- gulp\n- libgdx\n- uikit\n- synchronization\n- timezone\n- azure-web-app-service\n- dom-events\n- wso2\n- xampp\n- google-sheets-formula\n- crystal-reports\n- aggregation-framework\n- namespaces\n- swagger\n- android-emulator\n- uiscrollview\n- jvm\n- sequelize.js\n- chart.js\n- com\n- snowflake-cloud-data-platform\n- subprocess\n- geolocation\n- webdriver\n- garbage-collection\n- html5-canvas\n- sql-update\n- centos\n- dialog\n- concatenation\n- numbers\n- widget\n- qml\n- tuples\n- set\n- java-stream\n- mapreduce\n- ionic2\n- smtp\n- windows-10\n- android-edittext\n- nuget\n- rotation\n- spring-data\n- modal-dialog\n- radio-button\n- doctrine\n- http-headers\n- grid\n- lucene\n- sonarqube\n- xmlhttprequest\n- listbox\n- switch-statement\n- initialization\n- internationalization\n- components\n- boolean\n- apache-camel\n- google-play\n- gdb\n- serial-port\n- ios5\n- ldap\n- return\n- youtube-api\n- pivot\n- eclipse-plugin\n- latex\n- frameworks\n- tags\n- containers\n- c++17\n- subquery\n- github-actions\n- embedded\n- dataset\n- asp-classic\n- foreign-keys\n- label\n- uinavigationcontroller\n- delegates\n- copy\n- google-cloud-storage\n- struts2\n- migration\n- protractor\n- base64\n- uibutton\n- queue\n- find\n- sql-server-2008-r2\n- arguments\n- composer-php\n- append\n- jaxb\n- stack\n- zip\n- tailwind-css\n- cucumber\n- autolayout\n- ide\n- entity-framework-6\n- iteration\n- popup\n- r-markdown\n- windows-7\n- vb6\n- airflow\n- g++\n- clang\n- hover\n- ssl-certificate\n- jqgrid\n- range\n- gmail\n\nNext\n\nYou’ll be prompted to create an account to view your personalized homepage.\n\n1. 1. [Home](https:\/\/stackoverflow.com\/)\n   2. [Questions](https:\/\/stackoverflow.com\/questions)\n   3. [AI Assist](https:\/\/stackoverflow.com\/ai-assist)\n   4. [Tags](https:\/\/stackoverflow.com\/tags)\n   5. [Challenges](https:\/\/stackoverflow.com\/beta\/challenges)\n   6. [Chat](https:\/\/chat.stackoverflow.com\/?tab=explore)\n   7. [Articles](https:\/\/stackoverflow.blog\/contributed?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=so-blog&utm_content=experiment-articles)\n   8. [Users](https:\/\/stackoverflow.com\/users)\n   9. [Companies](https:\/\/stackoverflow.com\/jobs\/companies?so_medium=stackoverflow&so_source=SiteNav)\n   10. [Collectives]()\n   11. Communities for your favorite technologies. [Explore all Collectives](https:\/\/stackoverflow.com\/collectives-all)\n2. Stack Internal\n   Stack Overflow for Teams is now called **Stack Internal**. Bring the best of human thought and AI automation together at your work.\n   \n   [Try for free](https:\/\/stackoverflowteams.com\/teams\/create\/free\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=side-bar&utm_content=explore-teams) [Learn more](https:\/\/stackoverflow.co\/internal\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=side-bar&utm_content=explore-teams)\n3. [Stack Internal]()\n4. Bring the best of human thought and AI automation together at your work. [Learn more](https:\/\/stackoverflow.co\/internal\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=side-bar&utm_content=explore-teams-compact)\n\n##### Collectives™ on Stack Overflow\nFind centralized, trusted content and collaborate around the technologies you use most.\n\n[Learn more about Collectives](https:\/\/stackoverflow.com\/collectives)\n\n**Stack Internal**\n\nKnowledge at work\n\nBring the best of human thought and AI automation together at your work.\n\n[Explore Stack Internal](https:\/\/stackoverflow.co\/internal\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=side-bar&utm_content=explore-teams-compact-popover)\n\n# [How does one use Pytorch (+ cuda) with an A100 GPU?](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu)\n[Ask Question](https:\/\/stackoverflow.com\/questions\/ask)\n\nAsked\n\n4 years, 9 months ago\n\nModified [7 months ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu?lastactivity \"2025-06-18 09:01:10Z\")\n\nViewed 75k times\n\nThis question shows research effort; it is useful and clear\n\n47\n\nSave this question.\n\nShow activity on this post.\n\nI was trying to use my current code with an A100 gpu but I get this error:\n```\nCopy\n```\nwhich is reather confusing because it points to the usual pytorch installation but doesn't tell me which combination of pytorch version + cuda version to use for my specific hardware (A100). What is the right way to install pytorch for an A100?\n***\nThese are some versions I've tried:\n```\nCopy\n```\n***\nnote that this can be subtle because I've had this error with this machine + pytorch version in the past:\n\n[How to solve the famous \\`unhandled cuda error, NCCL version 2.7.8\\` error?](https:\/\/stackoverflow.com\/questions\/66807131\/how-to-solve-the-famous-unhandled-cuda-error-nccl-version-2-7-8-error)\n***\n# Bonus 1:\nI still have errors:\n```\nCopy\n```\none of the answers suggested to have nvcca & pytorch.version.cuda to match but they do not:\n```\nCopy\n```\nHow do I match them? I this the error? Can someone display their pip, conda and nvcca version to see what set up works?\n\nMore error messages:\n```\nCopy\n```\nafter putting\n```\nCopy\n```\n\n- [python](https:\/\/stackoverflow.com\/questions\/tagged\/python \"show questions tagged 'python'\")\n- [machine-learning](https:\/\/stackoverflow.com\/questions\/tagged\/machine-learning \"show questions tagged 'machine-learning'\")\n- [neural-network](https:\/\/stackoverflow.com\/questions\/tagged\/neural-network \"show questions tagged 'neural-network'\")\n- [pytorch](https:\/\/stackoverflow.com\/questions\/tagged\/pytorch \"show questions tagged 'pytorch'\")\n\n[Share](https:\/\/stackoverflow.com\/q\/66992585 \"Short permalink to this question\")\n\nShare a link to this question\n\nCopy link\n\n[CC BY-SA 4.0](https:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/ \"The current license for this post: CC BY-SA 4.0\")\n\n[Improve this question](https:\/\/stackoverflow.com\/posts\/66992585\/edit)\n\nFollow\n\nFollow this question to receive notifications\n\n[edited May 12, 2022 at 21:06](https:\/\/stackoverflow.com\/posts\/66992585\/revisions \"show all edits to this post\")\n\nasked Apr 7, 2021 at 19:04\n\n[![Charlie Parker's user avatar](https:\/\/i.sstatic.net\/2B6rV.png?s=64)](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\n[Charlie Parker](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\n6,2747979 gold badges264264 silver badges422422 bronze badges\n\n3\n\n- 2\n  Pytorch 1.7.0 or later with CUDA 11.0 or later [should work](https:\/\/www.marktechpost.com\/2020\/11\/01\/pytorch-releases-version-1-7-with-new-features-like-cuda-11-new-apis-for-ffts-and-nvidia-a100-generation-gpus-support\/). Or you could use [NGC](https:\/\/ngc.nvidia.com\/)\n  \n  Robert Crovella\n  \n  – [Robert Crovella](https:\/\/stackoverflow.com\/users\/1695960\/robert-crovella \"154,444 reputation\")\n  \n  2021-04-07 19:16:23 +00:00\n  \n  [Commented Apr 7, 2021 at 19:16](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment118418045_66992585)\n- 1\n  @RobertCrovella if what you say it's true then the command needed is `conda install -y pytorch==1.7.1 torchvision torchaudio cudatoolkit=11.0 -c pytorch -c conda-forge` will try soon if it worked.\n  \n  Charlie Parker\n  \n  – [Charlie Parker](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker \"6,274 reputation\")\n  \n  2021-04-24 01:38:45 +00:00\n  \n  [Commented Apr 24, 2021 at 1:38](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment118849441_66992585)\n- @CharlieParker just 14 minutes for the expiration of Bounty. non of these answers helpful?\n  \n  Sadra\n  \n  – [Sadra](https:\/\/stackoverflow.com\/users\/1076264\/sadra \"2,862 reputation\")\n  \n  2022-05-19 20:40:24 +00:00\n  \n  [Commented May 19, 2022 at 20:40](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127748943_66992585)\n\n[Add a comment](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu \"Use comments to ask for more information or suggest improvements. Avoid answering questions in comments.\")   \\|\n\n## 6 Answers 6\nSorted by:\n\n[Reset to default](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu?answertab=scoredesc#tab-top)\n\nThis answer is useful\n\n48\n\nSave this answer.\n\nShow activity on this post.\n\nFrom the link [pytorch site](https:\/\/pytorch.org\/get-started\/locally\/) from @SimonB 's answer, I did:\n```\nCopypip3 install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 -f https:\/\/download.pytorch.org\/whl\/torch_stable.html\n```\nThis solved the problem for me.\n\n[Share](https:\/\/stackoverflow.com\/a\/68626103 \"Short permalink to this answer\")\n\nShare a link to this answer\n\nCopy link\n\n[CC BY-SA 4.0](https:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/ \"The current license for this post: CC BY-SA 4.0\")\n\n[Improve this answer](https:\/\/stackoverflow.com\/posts\/68626103\/edit)\n\nFollow\n\nFollow this answer to receive notifications\n\nanswered Aug 2, 2021 at 18:02\n\n[![James Hirschorn's user avatar](https:\/\/www.gravatar.com\/avatar\/e4ee321232e2b4700138d891f6d2f004?s=64&d=identicon&r=PG)](https:\/\/stackoverflow.com\/users\/1349673\/james-hirschorn)\n\n[James Hirschorn](https:\/\/stackoverflow.com\/users\/1349673\/james-hirschorn)\n\n8,14466 gold badges5656 silver badges5858 bronze badges\n\nSign up to request clarification or add additional context in comments.\n\n## 6 Comments\nAdd a comment\n\n[![](https:\/\/www.gravatar.com\/avatar\/fed3f0aad2fe6040187cd2a7c10e7dcf?s=48&d=identicon&r=PG&f=y&so-version=2)](https:\/\/stackoverflow.com\/users\/7998544\/woma)\n\nWoma\n\n[Woma](https:\/\/stackoverflow.com\/users\/7998544\/woma)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment121805600_68626103)\n\nFor me, the conda installation did not work but the pip installation, no idea why\n\n2021-08-25T11:32:10.157Z+00:00\n\n1\n\nReply\n\n- Copy link\n\n[![](https:\/\/www.gravatar.com\/avatar\/aea0b2c6b6b44c5c7fe64f9a8561ffc8?s=48&d=identicon&r=PG&f=y&so-version=2)](https:\/\/stackoverflow.com\/users\/3188690\/tian)\n\nTian\n\n[Tian](https:\/\/stackoverflow.com\/users\/3188690\/tian)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment125318618_68626103)\n\nSomehow, it does not work for now, seems that the download link not work for some reason: \"returned a non-zero code: 137\"\n\n2022-01-28T01:53:45.39Z+00:00\n\n0\n\nReply\n\n- Copy link\n\n[![](https:\/\/i.sstatic.net\/2B6rV.png?s=64)](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\nCharlie Parker\n\n[Charlie Parker](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127600966_68626103)\n\nare you sure this is right? I have that my pytorch wants 11.1 but nvcca is 11.0 see: `(meta_learning_a100) [miranda9@hal-dgx ~]$ python -c \"import torch;print(torch.version.cuda)\" 11.1 (meta_learning_a100) [miranda9@hal-dgx ~]$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2020 NVIDIA Corporation Built on Wed_Jul_22_19:09:09_PDT_2020 Cuda compilation tools, release 11.0, V11.0.221 Build cuda_11.0_bu.TC445_37.28845127_0`\n\n2022-05-12T20:43:29.293Z+00:00\n\n0\n\nReply\n\n- Copy link\n\n[![](https:\/\/www.gravatar.com\/avatar\/e4ee321232e2b4700138d891f6d2f004?s=48&d=identicon&r=PG)](https:\/\/stackoverflow.com\/users\/1349673\/james-hirschorn)\n\nJames Hirschorn\n\n[James Hirschorn](https:\/\/stackoverflow.com\/users\/1349673\/james-hirschorn)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127668237_68626103)\n\n@CharlieParker It's been too long for me to recall the context for this question. But `nvidia-smi` reveals that I have CUDA 11.4 and nvcc 10.1\n\n2022-05-16T16:26:26.62Z+00:00\n\n0\n\nReply\n\n- Copy link\n\n[![](https:\/\/www.gravatar.com\/avatar\/3b495f8e4707e59490dda3cd0940c5e9?s=48&d=identicon&r=PG&f=y&so-version=2)](https:\/\/stackoverflow.com\/users\/8183621\/pascaliv)\n\nPascalIv\n\n[PascalIv](https:\/\/stackoverflow.com\/users\/8183621\/pascaliv)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment134468011_68626103)\n\nERROR: No matching distribution found for torch==1.9.0+cu111\n\n2023-05-15T09:12:04.423Z+00:00\n\n2\n\nReply\n\n- Copy link\n\nAdd a comment\n\n\\|\n\nShow 1 more comment\n\nThis answer is useful\n\n8\n\nSave this answer.\n\nShow activity on this post.\n\nI've got an A100 and have had success with\n```\nCopyconda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c nvidia\n```\nWhich is now also recommended on the [pytorch site](https:\/\/pytorch.org\/get-started\/locally\/)\n\n[Share](https:\/\/stackoverflow.com\/a\/67521477 \"Short permalink to this answer\")\n\nShare a link to this answer\n\nCopy link\n\n[CC BY-SA 4.0](https:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/ \"The current license for this post: CC BY-SA 4.0\")\n\n[Improve this answer](https:\/\/stackoverflow.com\/posts\/67521477\/edit)\n\nFollow\n\nFollow this answer to receive notifications\n\nanswered May 13, 2021 at 14:58\n\n[![Simon B's user avatar](https:\/\/www.gravatar.com\/avatar\/4a035074530513cee4b47303f64f904a?s=64&d=identicon&r=PG)](https:\/\/stackoverflow.com\/users\/1264592\/simon-b)\n\n[Simon B](https:\/\/stackoverflow.com\/users\/1264592\/simon-b)\n\n26711 silver badge77 bronze badges\n\n## 1 Comment\nAdd a comment\n\n[![](https:\/\/i.sstatic.net\/2B6rV.png?s=64)](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\nCharlie Parker\n\n[Charlie Parker](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127600972_67521477)\n\nare you sure this is right? I have that my pytorch wants 11.1 but nvcca is 11.0 see: `(meta_learning_a100) [miranda9@hal-dgx ~]$ python -c \"import torch;print(torch.version.cuda)\" 11.1 (meta_learning_a100) [miranda9@hal-dgx ~]$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2020 NVIDIA Corporation Built on Wed_Jul_22_19:09:09_PDT_2020 Cuda compilation tools, release 11.0, V11.0.221 Build cuda_11.0_bu.TC445_37.28845127_0`\n\n2022-05-12T20:43:54.453Z+00:00\n\n0\n\nReply\n\n- Copy link\n\nThis answer is useful\n\n4\n\nSave this answer.\n\nShow activity on this post.\n\nTo me this is what worked:\n```\nCopy\n```\nthen I tested it, asked for the device and did a matrix multiply, no errors is it worked:\n```\nCopy\n```\ngpu pytorch code:\n```\nCopy\n```\n\n[Share](https:\/\/stackoverflow.com\/a\/70946283 \"Short permalink to this answer\")\n\nShare a link to this answer\n\nCopy link\n\n[CC BY-SA 4.0](https:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/ \"The current license for this post: CC BY-SA 4.0\")\n\n[Improve this answer](https:\/\/stackoverflow.com\/posts\/70946283\/edit)\n\nFollow\n\nFollow this answer to receive notifications\n\nanswered Feb 1, 2022 at 19:18\n\n[![Charlie Parker's user avatar](https:\/\/i.sstatic.net\/2B6rV.png?s=64)](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\n[Charlie Parker](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\n6,2747979 gold badges264264 silver badges422422 bronze badges\n\n## 1 Comment\nAdd a comment\n\n[![](https:\/\/lh4.googleusercontent.com\/-6pGlNM6cH_4\/AAAAAAAAAAI\/AAAAAAAAACU\/YixjVAUWv20\/s48-rj\/photo.jpg)](https:\/\/stackoverflow.com\/users\/9058482\/harsh-kumar-chourasia)\n\nHarsh Kumar Chourasia\n\n[Harsh Kumar Chourasia](https:\/\/stackoverflow.com\/users\/9058482\/harsh-kumar-chourasia)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment129644183_70946283)\n\nThanks! This solution worked for me. I was trying to run pytorch inside docker. I uninstalled all the cuda libraries and pre-installed torch before running this command.\n\n2022-08-19T05:48:05.663Z+00:00\n\n0\n\nReply\n\n- Copy link\n\nThis answer is useful\n\n2\n\nSave this answer.\n\nShow activity on this post.\n\nI had the same problem. You need to install CUDA 11.0 instead of 10.2 and reinstall PyTorch for this CUDA version.\n\n[Share](https:\/\/stackoverflow.com\/a\/67026338 \"Short permalink to this answer\")\n\nShare a link to this answer\n\nCopy link\n\n[CC BY-SA 4.0](https:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/ \"The current license for this post: CC BY-SA 4.0\")\n\n[Improve this answer](https:\/\/stackoverflow.com\/posts\/67026338\/edit)\n\nFollow\n\nFollow this answer to receive notifications\n\nanswered Apr 9, 2021 at 18:15\n\n[![guillaumefrd's user avatar](https:\/\/lh3.googleusercontent.com\/-Q2BzoBitSPg\/AAAAAAAAAAI\/AAAAAAAAX1U\/mQloaeFMW24\/s64-rj\/photo.jpg)](https:\/\/stackoverflow.com\/users\/8926322\/guillaumefrd)\n\n[guillaumefrd](https:\/\/stackoverflow.com\/users\/8926322\/guillaumefrd)\n\n12766 bronze badges\n\n## 3 Comments\nAdd a comment\n\n[![](https:\/\/i.sstatic.net\/2B6rV.png?s=64)](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\nCharlie Parker\n\n[Charlie Parker](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment118848956_67026338)\n\ndid you install pytorch 1.8.0 using cuda 11.0 or pytorch 1.7.x?\n\n2021-04-24T00:40:10.057Z+00:00\n\n0\n\nReply\n\n- Copy link\n\n[![](https:\/\/lh3.googleusercontent.com\/-Q2BzoBitSPg\/AAAAAAAAAAI\/AAAAAAAAX1U\/mQloaeFMW24\/s48-rj\/photo.jpg)](https:\/\/stackoverflow.com\/users\/8926322\/guillaumefrd)\n\nguillaumefrd\n\n[guillaumefrd](https:\/\/stackoverflow.com\/users\/8926322\/guillaumefrd)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment118869529_67026338)\n\nI tried 1.8.0 and 1.7.1, both were working.\n\n2021-04-25T06:40:29.923Z+00:00\n\n1\n\nReply\n\n- Copy link\n\n[![](https:\/\/i.sstatic.net\/2B6rV.png?s=64)](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\nCharlie Parker\n\n[Charlie Parker](https:\/\/stackoverflow.com\/users\/1601580\/charlie-parker)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127600974_67026338)\n\nare you sure this is right? I have that my pytorch wants 11.1 but nvcca is 11.0 see: `(meta_learning_a100) [miranda9@hal-dgx ~]$ python -c \"import torch;print(torch.version.cuda)\" 11.1 (meta_learning_a100) [miranda9@hal-dgx ~]$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2020 NVIDIA Corporation Built on Wed_Jul_22_19:09:09_PDT_2020 Cuda compilation tools, release 11.0, V11.0.221 Build cuda_11.0_bu.TC445_37.28845127_0`\n\n2022-05-12T20:44:09.713Z+00:00\n\n0\n\nReply\n\n- Copy link\n\nAdd a comment\n\nThis answer is useful\n\n2\n\nSave this answer.\n\nShow activity on this post.\n\nThis solution is tested on a multi GPU **A100** environment:\n\ncreate a clean conda environment: `conda create -n pya100 python=3.9`\n\nthen check your nvcc version by: `nvcc --version` \\#mine return `11.3`\n\nthen install pytorch in this way: (as of now it installs Pytorch 1.11.0, torchvision 0.12.0)\n```\nCopyconda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch -c nvidia\n```\nnow `python -c \"import torch;print(torch.version.cuda)\"` returns `11.3` (though I don't think it matters that much)\n\nI shared my environment file [Here](https:\/\/pastebin.com\/qTRzFHjZ). You can build one environment based on using this: (just replace NAMEOFENVIRONMENT with your environment name)\n```\nCopyconda env update --name NAMEOFENVIRONMENT --file environment.yml\n```\n\n[Share](https:\/\/stackoverflow.com\/a\/72244310 \"Short permalink to this answer\")\n\nShare a link to this answer\n\nCopy link\n\n[CC BY-SA 4.0](https:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/ \"The current license for this post: CC BY-SA 4.0\")\n\n[Improve this answer](https:\/\/stackoverflow.com\/posts\/72244310\/edit)\n\nFollow\n\nFollow this answer to receive notifications\n\n[edited May 18, 2022 at 2:24](https:\/\/stackoverflow.com\/posts\/72244310\/revisions \"show all edits to this post\")\n\nanswered May 14, 2022 at 22:34\n\n[![Sadra's user avatar](https:\/\/i.sstatic.net\/M7tOf.jpg?s=64)](https:\/\/stackoverflow.com\/users\/1076264\/sadra)\n\n[Sadra](https:\/\/stackoverflow.com\/users\/1076264\/sadra)\n\n2,86222 gold badges2525 silver badges3434 bronze badges\n\n## 3 Comments\nAdd a comment\n\n[![](https:\/\/i.sstatic.net\/M7tOf.jpg?s=64)](https:\/\/stackoverflow.com\/users\/1076264\/sadra)\n\nSadra\n\n[Sadra](https:\/\/stackoverflow.com\/users\/1076264\/sadra)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127638212_72244310)\n\n@CharlieParker I think this address your problem\n\n2022-05-14T22:35:27.917Z+00:00\n\n0\n\nReply\n\n- Copy link\n\n[![](https:\/\/i.sstatic.net\/M7tOf.jpg?s=64)](https:\/\/stackoverflow.com\/users\/1076264\/sadra)\n\nSadra\n\n[Sadra](https:\/\/stackoverflow.com\/users\/1076264\/sadra)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127648326_72244310)\n\nif the current solution did not work please share `apt list --installed` through pastebin so I can compare if mine and give it another try\n\n2022-05-15T17:20:42.75Z+00:00\n\n0\n\nReply\n\n- Copy link\n\n[![](https:\/\/i.sstatic.net\/M7tOf.jpg?s=64)](https:\/\/stackoverflow.com\/users\/1076264\/sadra)\n\nSadra\n\n[Sadra](https:\/\/stackoverflow.com\/users\/1076264\/sadra)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127748922_72244310)\n\n@CharlieParker does it help?\n\n2022-05-19T20:39:31.213Z+00:00\n\n0\n\nReply\n\n- Copy link\n\nAdd a comment\n\nThis answer is useful\n\n1\n\nSave this answer.\n\nShow activity on this post.\n\nCheck your installed version of torch, torchvision, torchaudio etc. using\n```\nCopy<your virtualenv path>\/bin\/python -m torch.utils.collect_env\n```\nIn my case I had this -\n```\nCopy\n```\nSince, I was not using torchvision or torchaudio, I just updated my `torch` version using the suggestion by @JamesHirschorn and selected the one according to my `torch` version from this [pytorch](https:\/\/download.pytorch.org\/whl\/torch_stable.html) link. e.g. in my case, the `torch` version was 1.11.0 and hence I installed `torch==1.11.0+cu113`\n\n`pip install torch==1.11.0+cu113 -f https:\/\/download.pytorch.org\/whl\/torch_stable.html`\n\nAfter update, the output of `<your virtualenv path>\/bin\/python -m torch.utils.collect_env` was\n```\nCopy\n```\n\n[Share](https:\/\/stackoverflow.com\/a\/72106714 \"Short permalink to this answer\")\n\nShare a link to this answer\n\nCopy link\n\n[CC BY-SA 4.0](https:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/ \"The current license for this post: CC BY-SA 4.0\")\n\n[Improve this answer](https:\/\/stackoverflow.com\/posts\/72106714\/edit)\n\nFollow\n\nFollow this answer to receive notifications\n\nanswered May 4, 2022 at 0:21\n\n[![rhn89's user avatar](https:\/\/i.sstatic.net\/HDWUs.jpg?s=64)](https:\/\/stackoverflow.com\/users\/3082461\/rhn89)\n\n[rhn89](https:\/\/stackoverflow.com\/users\/3082461\/rhn89)\n\n44855 silver badges1111 bronze badges\n\n## 1 Comment\nAdd a comment\n\n[![](https:\/\/i.sstatic.net\/6rTV9.png?s=64)](https:\/\/stackoverflow.com\/users\/979474\/user956584)\n\nuser956584\n\n[user956584](https:\/\/stackoverflow.com\/users\/979474\/user956584)\n\n[Over a year ago](https:\/\/stackoverflow.com\/questions\/66992585\/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127424655_72106714)\n\nfor me with message: Tesla K20Xm with CUDA capability sm\\_35 is not compatible with the current PyTorch installation. this works: pip install torch==1.2.0+cu92 -f [download.pytorch.org\/whl\/torch\\_stable.html](https:\/\/download.pytorch.org\/whl\/torch_stable.html)\n\n2022-05-04T18:03:05.943Z+00:00\n\n1\n\nReply\n\n- Copy link\n\nStart asking to get answers\n\nFind the answer to your question by asking.\n\n[Ask question](https:\/\/stackoverflow.com\/questions\/ask)\n\nExplore related questions\n\n- [python](https:\/\/stackoverflow.com\/questions\/tagged\/python \"show questions tagged 'python'\")\n- [machine-learning](https:\/\/stackoverflow.com\/questions\/tagged\/machine-learning \"show questions tagged 'machine-learning'\")\n- [neural-network](https:\/\/stackoverflow.com\/questions\/tagged\/neural-network \"show questions tagged 'neural-network'\")\n- [pytorch](https:\/\/stackoverflow.com\/questions\/tagged\/pytorch \"show questions tagged 'pytorch'\")\n\nSee similar questions with these tags.\n\n- The Overflow Blog\n- [Is that allowed? Authentication and authorization in Model Context Protocol](https:\/\/stackoverflow.blog\/2026\/01\/21\/is-that-allowed-authentication-and-authorization-in-model-context-protocol\/?cb=1)\n- [Automate your security whack-a-mole: Q\\&A with Exaforce](https:\/\/stackoverflow.blog\/2026\/01\/22\/automate-your-security-whack-a-mole-q-and-a-with-exaforce\/?cb=1)\n- Featured on Meta\n- [AI Assist: recent updates (January 2026)](https:\/\/meta.stackexchange.com\/questions\/416954\/ai-assist-recent-updates-january-2026?cb=1)\n- [Community Engagement Across the Network: Focus for 2026](https:\/\/meta.stackexchange.com\/questions\/416985\/community-engagement-across-the-network-focus-for-2026?cb=1)\n- [All users on Stack Overflow can now participate in chat](https:\/\/meta.stackoverflow.com\/questions\/437932\/all-users-on-stack-overflow-can-now-participate-in-chat?cb=1)\n- [Policy: Generative AI (e.g., ChatGPT) is banned](https:\/\/meta.stackoverflow.com\/questions\/421831\/policy-generative-ai-e-g-chatgpt-is-banned?cb=1)\n- [Stack Overflow now uses machine learning to flag spam automatically](https:\/\/meta.stackoverflow.com\/questions\/437959\/stack-overflow-now-uses-machine-learning-to-flag-spam-automatically?cb=1)\n- [No, I do not believe this is the end](https:\/\/meta.stackoverflow.com\/questions\/438009\/no-i-do-not-believe-this-is-the-end?cb=1)\n\nCommunity activity\n\nLast 1 hr\n\n- Users online activity\n  5257 users online\n- 11 questions\n- 16 answers\n- 26 comments\n- 156 upvotes\n\nPopular tags\n\n[image-processing](https:\/\/stackoverflow.com\/questions\/tagged\/image-processing)[excel](https:\/\/stackoverflow.com\/questions\/tagged\/excel)[png](https:\/\/stackoverflow.com\/questions\/tagged\/png)[html](https:\/\/stackoverflow.com\/questions\/tagged\/html)[client-side](https:\/\/stackoverflow.com\/questions\/tagged\/client-side)[python](https:\/\/stackoverflow.com\/questions\/tagged\/python)\n\nPopular unanswered question\n\n[MongoDB querySrv ECONNREFUSED on Node.js 22 after switching Node versions (worked before, now fails)](https:\/\/stackoverflow.com\/questions\/79873598)\n\n[node.js](https:\/\/stackoverflow.com\/questions\/tagged\/node.js)[mongodb](https:\/\/stackoverflow.com\/questions\/tagged\/mongodb)[windows](https:\/\/stackoverflow.com\/questions\/tagged\/windows)[express](https:\/\/stackoverflow.com\/questions\/tagged\/express)[mongoose](https:\/\/stackoverflow.com\/questions\/tagged\/mongoose)\n\n[![User avatar](https:\/\/www.gravatar.com\/avatar\/66dea90129372b0e057d203049c3142b?s=256&d=identicon&r=PG&f=y&so-version=2)](https:\/\/stackoverflow.com\/users\/32237460)\n\n[Vin](https:\/\/stackoverflow.com\/users\/32237460)\n\n- 11\n\n19 hours ago\n\n#### Linked\n[23](https:\/\/stackoverflow.com\/questions\/66807131\/how-to-solve-the-famous-unhandled-cuda-error-nccl-version-2-7-8-error?lq=1 \"Question score (upvotes - downvotes)\")\n\n[How to solve the famous \\`unhandled cuda error, NCCL version 2.7.8\\` error?](https:\/\/stackoverflow.com\/questions\/66807131\/how-to-solve-the-famous-unhandled-cuda-error-nccl-version-2-7-8-error?noredirect=1&lq=1)\n\n[0](https:\/\/stackoverflow.com\/questions\/76366411\/install-pytorch-dependencies-torchtext-torchdata-torch-with-cuda-and-a100?lq=1 \"Question score (upvotes - downvotes)\")\n\n[Install Pytorch dependencies (torchtext + torchdata + torch) with cuda and A100 GPU](https:\/\/stackoverflow.com\/questions\/76366411\/install-pytorch-dependencies-torchtext-torchdata-torch-with-cuda-and-a100?noredirect=1&lq=1)\n\n#### Related\n[1](https:\/\/stackoverflow.com\/questions\/50771001\/pytorch-not-using-cuda-device?rq=3 \"Question score (upvotes - downvotes)\")\n\n[Pytorch not using cuda device](https:\/\/stackoverflow.com\/questions\/50771001\/pytorch-not-using-cuda-device?rq=3)\n\n[46](https:\/\/stackoverflow.com\/questions\/50954479\/using-cuda-with-pytorch?rq=3 \"Question score (upvotes - downvotes)\")\n\n[Using CUDA with pytorch?](https:\/\/stackoverflow.com\/questions\/50954479\/using-cuda-with-pytorch?rq=3)\n\n[12](https:\/\/stackoverflow.com\/questions\/51605893\/why-doesnt-my-simple-pytorch-network-work-on-gpu-device?rq=3 \"Question score (upvotes - downvotes)\")\n\n[Why doesn't my simple pytorch network work on GPU device?](https:\/\/stackoverflow.com\/questions\/51605893\/why-doesnt-my-simple-pytorch-network-work-on-gpu-device?rq=3)\n\n[4](https:\/\/stackoverflow.com\/questions\/56852347\/model-cuda-in-pytorch?rq=3 \"Question score (upvotes - downvotes)\")\n\n[model.cuda() in pytorch](https:\/\/stackoverflow.com\/questions\/56852347\/model-cuda-in-pytorch?rq=3)\n\n[2](https:\/\/stackoverflow.com\/questions\/59497887\/how-can-i-convert-pytorch-cpu-based-transformation-to-cuda-based?rq=3 \"Question score (upvotes - downvotes)\")\n\n[How can I convert pytorch cpu-based transformation to cuda-based?](https:\/\/stackoverflow.com\/questions\/59497887\/how-can-i-convert-pytorch-cpu-based-transformation-to-cuda-based?rq=3)\n\n[0](https:\/\/stackoverflow.com\/questions\/60063910\/forcing-pytorch-to-use-gpu?rq=3 \"Question score (upvotes - downvotes)\")\n\n[Forcing Pytorch to use GPU](https:\/\/stackoverflow.com\/questions\/60063910\/forcing-pytorch-to-use-gpu?rq=3)\n\n[0](https:\/\/stackoverflow.com\/questions\/62476250\/how-to-run-a-pre-trained-pytorch-model-on-the-gpu?rq=3 \"Question score (upvotes - downvotes)\")\n\n[How to run a pre-trained pytorch model on the GPU?](https:\/\/stackoverflow.com\/questions\/62476250\/how-to-run-a-pre-trained-pytorch-model-on-the-gpu?rq=3)\n\n[25](https:\/\/stackoverflow.com\/questions\/63423463\/using-pytorch-cuda-on-macbook-pro?rq=3 \"Question score (upvotes - downvotes)\")\n\n[Using pytorch Cuda on MacBook Pro](https:\/\/stackoverflow.com\/questions\/63423463\/using-pytorch-cuda-on-macbook-pro?rq=3)\n\n[4](https:\/\/stackoverflow.com\/questions\/67645531\/how-does-one-run-pytorch-on-a-a40-gpu-without-errors-with-ddp-too?rq=3 \"Question score (upvotes - downvotes)\")\n\n[How does one run PyTorch on a A40 GPU without errors (with DDP too)?](https:\/\/stackoverflow.com\/questions\/67645531\/how-does-one-run-pytorch-on-a-a40-gpu-without-errors-with-ddp-too?rq=3)\n\n[0](https:\/\/stackoverflow.com\/questions\/69591391\/pytorch-using-cuda-prevents-optimization-from-working?rq=3 \"Question score (upvotes - downvotes)\")\n\n[Pytorch: using CUDA prevents optimization from working](https:\/\/stackoverflow.com\/questions\/69591391\/pytorch-using-cuda-prevents-optimization-from-working?rq=3)\n\n#### [Hot Network Questions](https:\/\/stackexchange.com\/questions?tab=hot)\n- [Question about Einstein problem](https:\/\/mathoverflow.net\/questions\/507294\/question-about-einstein-problem)\n- [Is there a deeper reason why surjections are harder to count than injections?](https:\/\/math.stackexchange.com\/questions\/5120661\/is-there-a-deeper-reason-why-surjections-are-harder-to-count-than-injections)\n- [Abandoned art and copyright?](https:\/\/law.stackexchange.com\/questions\/113979\/abandoned-art-and-copyright)\n- [Have there been major developments in the Langlands Program in the number field setting since 2010?](https:\/\/mathoverflow.net\/questions\/507242\/have-there-been-major-developments-in-the-langlands-program-in-the-number-field)\n- [JSON Parser in C++](https:\/\/codereview.stackexchange.com\/questions\/301080\/json-parser-in-c)\n- [Workbench on Spring'26 sandbox: OAUTH\\_APPROVAL\\_ERROR\\_GENERIC](https:\/\/salesforce.stackexchange.com\/questions\/438640\/workbench-on-spring26-sandbox-oauth-approval-error-generic)\n- [Is there any way of moving the symbol vertically wrt the notehead?](https:\/\/music.stackexchange.com\/questions\/143091\/is-there-any-way-of-moving-the-symbol-vertically-wrt-the-notehead)\n- [How do I stop the radidons beak from cracking?](https:\/\/worldbuilding.stackexchange.com\/questions\/272416\/how-do-i-stop-the-radidons-beak-from-cracking)\n- [How can I maintain tension in a PF2e, a system with unlimited out-of-combat healing?](https:\/\/rpg.stackexchange.com\/questions\/218576\/how-can-i-maintain-tension-in-a-pf2e-a-system-with-unlimited-out-of-combat-heal)\n- [Dual passports (UK\/HK) holder entering & leaving the US](https:\/\/travel.stackexchange.com\/questions\/203124\/dual-passports-uk-hk-holder-entering-leaving-the-us)\n- [\"Sed nescio quomodo nihil tam absurde dici potest quod non dicatur ab aliquo philosophorum.\" - Why \"quod\" and not \"ut\" or \"quam\"?](https:\/\/latin.stackexchange.com\/questions\/27062\/sed-nescio-quomodo-nihil-tam-absurde-dici-potest-quod-non-dicatur-ab-aliquo-phi)\n- [(I Cor 2:16) What actually \\*is\\* the \"mind of Christ\"? Is it the Spirit of knowledge, wisdom, understanding, counsel and might in Isa 11:2?](https:\/\/hermeneutics.stackexchange.com\/questions\/114138\/i-cor-216-what-actually-is-the-mind-of-christ-is-it-the-spirit-of-knowle)\n- [Looking for an old physics lab meme (early 2000s)](https:\/\/physics.stackexchange.com\/questions\/868126\/looking-for-an-old-physics-lab-meme-early-2000s)\n- [Batcher's sort in Java: a θ(N log N log N) sorting algorithm](https:\/\/codereview.stackexchange.com\/questions\/301081\/batchers-sort-in-java-a-%CE%B8n-log-n-log-n-sorting-algorithm)\n- [Arrange 11 (or fewer) squares that will be colored](https:\/\/puzzling.stackexchange.com\/questions\/136808\/arrange-11-or-fewer-squares-that-will-be-colored)\n- [What did John Austin (died 1859) mean by habit, habitual(ly)?](https:\/\/philosophy.stackexchange.com\/questions\/135453\/what-did-john-austin-died-1859-mean-by-habit-habitually)\n- [Coming-of-age short story involving a boy abducted by aliens](https:\/\/scifi.stackexchange.com\/questions\/302964\/coming-of-age-short-story-involving-a-boy-abducted-by-aliens)\n- [Approximating a polynomial with one with a particular Galois group](https:\/\/mathoverflow.net\/questions\/507277\/approximating-a-polynomial-with-one-with-a-particular-galois-group)\n- [GCC miscompiles long chains of overloaded \\`operator,\\`](https:\/\/stackoverflow.com\/questions\/79872387\/gcc-miscompiles-long-chains-of-overloaded-operator)\n- [Efficiency when converting camelCase to snake\\_case](https:\/\/codereview.stackexchange.com\/questions\/301067\/efficiency-when-converting-camelcase-to-snake-case)\n- [My direct manager accepts my timesheet, RM goes back in and takes away some of my hours](https:\/\/workplace.stackexchange.com\/questions\/203103\/my-direct-manager-accepts-my-timesheet-rm-goes-back-in-and-takes-away-some-of-m)\n- [Where are the words? A connect wall](https:\/\/puzzling.stackexchange.com\/questions\/136800\/where-are-the-words-a-connect-wall)\n- [Intuitive analysis of d'Alembert's solution to the 1+1D wave equation](https:\/\/physics.stackexchange.com\/questions\/868193\/intuitive-analysis-of-dalemberts-solution-to-the-11d-wave-equation)\n- [Construction of a large stealth spacecraft](https:\/\/worldbuilding.stackexchange.com\/questions\/272415\/construction-of-a-large-stealth-spacecraft)\n\n[Question feed](https:\/\/stackoverflow.com\/feeds\/question\/66992585 \"Feed of this question and its answers\")\n\n# Subscribe to RSS\nQuestion feed\n\nTo subscribe to this RSS feed, copy and paste this URL into your RSS reader.\n\n![](https:\/\/stackoverflow.com\/posts\/66992585\/ivc\/b2a4?prg=f420b4f9-cf6a-45f5-a265-d6a1b7a03d68)\n\nlang-py\n\n# Why are you flagging this comment?\nProbable spam.\n\nThis comment promotes a product, service or website while [failing to disclose the author's affiliation](https:\/\/stackoverflow.com\/help\/promotion).\n\nUnfriendly or contains harassment\/bigotry\/abuse.\n\nThis comment is unkind, insulting or attacks another person or group. Learn more in our [Abusive behavior policy](https:\/\/stackoverflow.com\/conduct\/abusive-behavior).\n\nNot needed.\n\nThis comment is not relevant to the post.\n\n```\n\n```\nEnter at least 6 characters\n\nSomething else.\n\nA problem not listed above. Try to be as specific as possible.\n\n```\n\n```\nEnter at least 6 characters\n\nFlag comment\n\nCancel\n\nYou have 0 flags left today\n\n![](https:\/\/stackoverflow.com\/js-false.gif)\n\n![](https:\/\/stackoverflow.com\/js-true.gif)\n\n# ![Illustration of upvote icon after it is clicked](https:\/\/stackoverflow.com\/Content\/Img\/modal\/img-upvote.png?v=fce73bd9724d)\n# Hang on, you can't upvote just yet.\nYou'll need to complete a few actions and gain 15 reputation points before being able to upvote. **Upvoting** indicates when questions and answers are useful. [What's reputation and how do I get it?](https:\/\/stackoverflow.com\/help\/whats-reputation)\n\nInstead, you can save this post to reference later.\n\nSave this post for later\n\nNot now\n\n##### [Stack Overflow](https:\/\/stackoverflow.com\/)\n- [Questions](https:\/\/stackoverflow.com\/questions)\n- [Help](https:\/\/stackoverflow.com\/help)\n- [Chat](https:\/\/chat.stackoverflow.com\/?tab=explore)\n\n##### [Business](https:\/\/stackoverflow.co\/)\n- [Stack Internal](https:\/\/stackoverflow.co\/internal\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=footer&utm_content=teams)\n- [Stack Data Licensing](https:\/\/stackoverflow.co\/data-licensing\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=footer&utm_content=data-licensing)\n- [Stack Ads](https:\/\/stackoverflow.co\/advertising\/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=footer&utm_content=advertising)\n\n##### [Company](https:\/\/stackoverflow.co\/)\n- [About](https:\/\/stackoverflow.co\/)\n- [Press](https:\/\/stackoverflow.co\/company\/press\/)\n- [Work Here](https:\/\/stackoverflow.co\/company\/work-here\/)\n- [Legal](https:\/\/stackoverflow.com\/legal)\n- [Privacy Policy](https:\/\/stackoverflow.com\/legal\/privacy-policy)\n- [Terms of Service](https:\/\/stackoverflow.com\/legal\/terms-of-service\/public)\n- [Contact Us](https:\/\/stackoverflow.com\/contact)\n- Cookie Settings\n- [Cookie Policy](https:\/\/policies.stackoverflow.co\/stack-overflow\/cookie-policy)\n\n##### [Stack Exchange Network](https:\/\/stackexchange.com\/)\n- [Technology](https:\/\/stackexchange.com\/sites#technology)\n- [Culture & recreation](https:\/\/stackexchange.com\/sites#culturerecreation)\n- [Life & arts](https:\/\/stackexchange.com\/sites#lifearts)\n- [Science](https:\/\/stackexchange.com\/sites#science)\n- [Professional](https:\/\/stackexchange.com\/sites#professional)\n- [Business](https:\/\/stackexchange.com\/sites#business)\n- [API](https:\/\/api.stackexchange.com\/)\n- [Data](https:\/\/data.stackexchange.com\/)\n\n- [Blog](https:\/\/stackoverflow.blog\/?blb=1)\n- [Facebook](https:\/\/www.facebook.com\/officialstackoverflow\/)\n- [Twitter](https:\/\/twitter.com\/stackoverflow)\n- [LinkedIn](https:\/\/linkedin.com\/company\/stack-overflow)\n- [Instagram](https:\/\/www.instagram.com\/thestackoverflow)\n\nSite design \/ logo © 2026 Stack Exchange Inc;  user contributions licensed under  [CC BY-SA](https:\/\/stackoverflow.com\/help\/licensing) .  rev 2026.1.22.38918","attrs_readable_markdown":null,"meta_canonical":null}

3. Robots.txt Check

Query:

Response:

4. Spam/Ban Check

Query:

Response:

5. Seen Status Check

ℹ️ Skipped - page is already crawled

📄

INDEXABLE

✅

CRAWLED

5 days ago

🤖

ROBOTS ALLOWED

Page Info Filters

Filter	Status	Condition	Details
HTTP status	PASS	`download_http_code = 200`	HTTP 200
Age cutoff	PASS	`download_stamp > now() - 6 MONTH`	0.2 months ago
History drop	PASS	`isNull(history_drop_reason)`	No drop reason
Spam/ban	PASS	`fh_dont_index != 1 AND ml_spam_score = 0`	ml_spam_score=0
Canonical	PASS	`meta_canonical IS NULL OR = '' OR = src_unparsed`	Not set

Page Details

Property	Value
URL	https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu
Last Crawled	2026-04-02 22:45:42 (5 days ago)
First Indexed	2021-09-04 12:36:50 (4 years ago)
HTTP Status Code	200
Meta Title	python - How does one use Pytorch (+ cuda) with an A100 GPU? - Stack Overflow
Meta Description	null
Meta Canonical	null
Boilerpipe Text	I was trying to use my current code with an A100 gpu but I get this error: ---> backend= 'nccl' /home/miranda9/miniconda3/envs/metalearningpy1 .7 .1 c10 .2 /lib/python3 .8 /site-packages/torch/cuda/__init__.py: 104 : UserWarning: A100-SXM4-40GB with CUDA capability sm_80 is not compatible with the current PyTorch installation. The current PyTorch install supports CUDA capabilities sm_37 sm_50 sm_60 sm_61 sm_70 sm_75 compute_37. If you want to use the A100-SXM4-40GB GPU with PyTorch, please check the instructions at https://pytorch.org/get-started/locally/ which is reather confusing because it points to the usual pytorch installation but doesn't tell me which combination of pytorch version + cuda version to use for my specific hardware (A100). What is the right way to install pytorch for an A100? These are some versions I've tried: # conda install -y pytorch==1.8.0 torchvision cudatoolkit=10.2 -c pytorch # conda install -y pytorch torchvision cudatoolkit=10.2 -c pytorch #conda install -y pytorch==1.7.1 torchvision torchaudio cudatoolkit=10.2 -c pytorch -c conda-forge # conda install -y pytorch==1.6.0 torchvision cudatoolkit=10.2 -c pytorch #conda install -y pytorch==1.7.1 torchvision torchaudio cudatoolkit=11.1 -c pytorch -c conda-forge # conda install pytorch torchvision torchaudio cudatoolkit=11.0 -c pytorch # conda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c conda-forge # conda install -y pytorch torchvision cudatoolkit=9.2 -c pytorch # For Nano, CC # conda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c conda-forge note that this can be subtle because I've had this error with this machine + pytorch version in the past: How to solve the famous `unhandled cuda error, NCCL version 2.7.8` error? Bonus 1: I still have errors: ncclSystemError: System call (socket, malloc, munmap, etc) failed. Traceback (most recent call last): File "/home/miranda9/diversity-for-predictive-success-of-meta-learning/div_src/diversity_src/experiment_mains/main_dist_maml_l2l.py" , line 1423 , in <module> main() File "/home/miranda9/diversity-for-predictive-success-of-meta-learning/div_src/diversity_src/experiment_mains/main_dist_maml_l2l.py" , line 1365 , in main train(args=args) File "/home/miranda9/diversity-for-predictive-success-of-meta-learning/div_src/diversity_src/experiment_mains/main_dist_maml_l2l.py" , line 1385 , in train args.opt = move_opt_to_cherry_opt_and_sync_params(args) if is_running_parallel(args.rank) else args.opt File "/home/miranda9/ultimate-utils/ultimate-utils-proj-src/uutils/torch_uu/distributed.py" , line 456 , in move_opt_to_cherry_opt_and_sync_params args.opt = cherry.optim.Distributed(args.model.parameters(), opt=args.opt, sync=syn) File "/home/miranda9/miniconda3/envs/meta_learning_a100/lib/python3.9/site-packages/cherry/optim.py" , line 62 , in __init__ self.sync_parameters() File "/home/miranda9/miniconda3/envs/meta_learning_a100/lib/python3.9/site-packages/cherry/optim.py" , line 78 , in sync_parameters dist.broadcast(p.data, src=root) File "/home/miranda9/miniconda3/envs/meta_learning_a100/lib/python3.9/site-packages/torch/distributed/distributed_c10d.py" , line 1090 , in broadcast work = default_pg.broadcast([tensor], opts) RuntimeError: NCCL error in : ../torch/lib/c10d/ProcessGroupNCCL.cpp: 911 , unhandled system error, NCCL version 2.7 .8 one of the answers suggested to have nvcca & pytorch.version.cuda to match but they do not: (meta_learning_a100) [miranda9@hal-dgx ~]$ python -c "import torch;print(torch.version.cuda)" 11.1 (meta_learning_a100) [miranda9@hal-dgx ~]$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005 - 2020 NVIDIA Corporation Built on Wed_Jul_22_19:09:09_PDT_2020 Cuda compilation tools, release 11.0 , V11 .0 .221 Build cuda_11 .0 _bu.TC445_37 .28845127_0 How do I match them? I this the error? Can someone display their pip, conda and nvcca version to see what set up works? More error messages: hal-dgx: 21797 : 21797 [ 0 ] NCCL INFO Bootstrap : Using [ 0 ]enp226s0: 141.142 .153 .83 < 0 > [ 1 ]virbr0: 192.168 .122 .1 < 0 > hal-dgx: 21797 : 21797 [ 0 ] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation hal-dgx: 21797 : 21797 [ 0 ] NCCL INFO NET/IB : Using [ 0 ]mlx5_0: 1 /IB [ 1 ]mlx5_1: 1 /IB [ 2 ]mlx5_2: 1 /IB [ 3 ]mlx5_3: 1 /IB [ 4 ]mlx5_4: 1 /IB [ 5 ]mlx5_5: 1 /IB [ 6 ]mlx5_6: 1 /IB [ 7 ]mlx5_7: 1 /IB ; OOB enp226s0: 141.142 .153 .83 < 0 > hal-dgx: 21797 : 21797 [ 0 ] NCCL INFO Using network IB NCCL version 2.7 .8 +cuda11 .1 hal-dgx: 21805 : 21805 [ 2 ] NCCL INFO Bootstrap : Using [ 0 ]enp226s0: 141.142 .153 .83 < 0 > [ 1 ]virbr0: 192.168 .122 .1 < 0 > hal-dgx: 21799 : 21799 [ 1 ] NCCL INFO Bootstrap : Using [ 0 ]enp226s0: 141.142 .153 .83 < 0 > [ 1 ]virbr0: 192.168 .122 .1 < 0 > hal-dgx: 21805 : 21805 [ 2 ] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation hal-dgx: 21799 : 21799 [ 1 ] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation hal-dgx: 21811 : 21811 [ 3 ] NCCL INFO Bootstrap : Using [ 0 ]enp226s0: 141.142 .153 .83 < 0 > [ 1 ]virbr0: 192.168 .122 .1 < 0 > hal-dgx: 21811 : 21811 [ 3 ] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation hal-dgx: 21811 : 21811 [ 3 ] NCCL INFO NET/IB : Using [ 0 ]mlx5_0: 1 /IB [ 1 ]mlx5_1: 1 /IB [ 2 ]mlx5_2: 1 /IB [ 3 ]mlx5_3: 1 /IB [ 4 ]mlx5_4: 1 /IB [ 5 ]mlx5_5: 1 /IB [ 6 ]mlx5_6: 1 /IB [ 7 ]mlx5_7: 1 /IB ; OOB enp226s0: 141.142 .153 .83 < 0 > hal-dgx: 21811 : 21811 [ 3 ] NCCL INFO Using network IB hal-dgx: 21799 : 21799 [ 1 ] NCCL INFO NET/IB : Using [ 0 ]mlx5_0: 1 /IB [ 1 ]mlx5_1: 1 /IB [ 2 ]mlx5_2: 1 /IB [ 3 ]mlx5_3: 1 /IB [ 4 ]mlx5_4: 1 /IB [ 5 ]mlx5_5: 1 /IB [ 6 ]mlx5_6: 1 /IB [ 7 ]mlx5_7: 1 /IB ; OOB enp226s0: 141.142 .153 .83 < 0 > hal-dgx: 21805 : 21805 [ 2 ] NCCL INFO NET/IB : Using [ 0 ]mlx5_0: 1 /IB [ 1 ]mlx5_1: 1 /IB [ 2 ]mlx5_2: 1 /IB [ 3 ]mlx5_3: 1 /IB [ 4 ]mlx5_4: 1 /IB [ 5 ]mlx5_5: 1 /IB [ 6 ]mlx5_6: 1 /IB [ 7 ]mlx5_7: 1 /IB ; OOB enp226s0: 141.142 .153 .83 < 0 > hal-dgx: 21799 : 21799 [ 1 ] NCCL INFO Using network IB hal-dgx: 21805 : 21805 [ 2 ] NCCL INFO Using network IB hal-dgx: 21797 : 27906 [ 0 ] misc/ibvwrap.cc: 280 NCCL WARN Call to ibv_create_qp failed hal-dgx: 21797 : 27906 [ 0 ] NCCL INFO transport/net_ib.cc: 360 -> 2 hal-dgx: 21797 : 27906 [ 0 ] NCCL INFO transport/net_ib.cc: 437 -> 2 hal-dgx: 21797 : 27906 [ 0 ] NCCL INFO include/net.h: 21 -> 2 hal-dgx: 21797 : 27906 [ 0 ] NCCL INFO include/net.h: 51 -> 2 hal-dgx: 21797 : 27906 [ 0 ] NCCL INFO init.cc: 300 -> 2 hal-dgx: 21797 : 27906 [ 0 ] NCCL INFO init.cc: 566 -> 2 hal-dgx: 21797 : 27906 [ 0 ] NCCL INFO init.cc: 840 -> 2 hal-dgx: 21797 : 27906 [ 0 ] NCCL INFO group.cc: 73 -> 2 [Async thread] hal-dgx: 21811 : 27929 [ 3 ] misc/ibvwrap.cc: 280 NCCL WARN Call to ibv_create_qp failed hal-dgx: 21811 : 27929 [ 3 ] NCCL INFO transport/net_ib.cc: 360 -> 2 hal-dgx: 21811 : 27929 [ 3 ] NCCL INFO transport/net_ib.cc: 437 -> 2 hal-dgx: 21811 : 27929 [ 3 ] NCCL INFO include/net.h: 21 -> 2 hal-dgx: 21811 : 27929 [ 3 ] NCCL INFO include/net.h: 51 -> 2 hal-dgx: 21811 : 27929 [ 3 ] NCCL INFO init.cc: 300 -> 2 hal-dgx: 21811 : 27929 [ 3 ] NCCL INFO init.cc: 566 -> 2 hal-dgx: 21811 : 27929 [ 3 ] NCCL INFO init.cc: 840 -> 2 hal-dgx: 21811 : 27929 [ 3 ] NCCL INFO group.cc: 73 -> 2 [Async thread] after putting import os os.environ[ "NCCL_DEBUG" ] = "INFO"
Markdown	# ![site logo](https://stackoverflow.com/Content/Img/SE-logo75.png) By clicking “Sign up”, you agree to our [terms of service](https://stackoverflow.com/legal/terms-of-service/public) and acknowledge you have read our [privacy policy](https://stackoverflow.com/legal/privacy-policy). # OR Already have an account? [Log in](https://stackoverflow.com/users/login) [Skip to main content](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#content) [Stack Overflow](https://stackoverflow.com/) 1. [About](https://stackoverflow.co/) 2. Products 3. [For Teams](https://stackoverflow.co/internal/) 1. [Stack Internal Implement a knowledge platform layer to power your enterprise and AI tools.](https://stackoverflow.co/internal/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=stack-overflow-for-teams) 2. [Stack Data Licensing Get access to top-class technical expertise with trusted & attributed content.](https://stackoverflow.co/data-licensing/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=overflow-api) 3. [Stack Ads Connect your brand to the world’s most trusted technologist communities.](https://stackoverflow.co/advertising/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=stack-overflow-advertising) 4. [Releases Keep up-to-date on features we add to Stack Overflow and Stack Internal.](https://stackoverflow.blog/releases/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=releases) 5. [About the company](https://stackoverflow.co/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=about-the-company) [Visit the blog](https://stackoverflow.blog/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=top-nav&utm_content=blog) 1. ### [current community](https://stackoverflow.com/) - [Stack Overflow](https://stackoverflow.com/) [help](https://stackoverflow.com/help) [chat](https://chat.stackoverflow.com/?tab=explore) - [Meta Stack Overflow](https://meta.stackoverflow.com/) ### your communities [Sign up](https://stackoverflow.com/users/signup?ssrc=site_switcher&returnurl=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F66992585%2Fhow-does-one-use-pytorch-cuda-with-an-a100-gpu) or [log in](https://stackoverflow.com/users/login?ssrc=site_switcher&returnurl=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F66992585%2Fhow-does-one-use-pytorch-cuda-with-an-a100-gpu) to customize your list. ### [more stack exchange communities](https://stackexchange.com/sites) [company blog](https://stackoverflow.blog/) 2. [Log in](https://stackoverflow.com/users/login?ssrc=head&returnurl=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F66992585%2Fhow-does-one-use-pytorch-cuda-with-an-a100-gpu) 3. [Sign up](https://stackoverflow.com/users/signup?ssrc=head&returnurl=https%3A%2F%2Fstackoverflow.com%2Fquestions%2F66992585%2Fhow-does-one-use-pytorch-cuda-with-an-a100-gpu) # Let's set up your homepage Select a few topics you're interested in: python javascript c\# reactjs java android html flutter c++ node.js typescript css r php angular next.js spring-boot machine-learning sql excel ios azure docker Or search from our full list: - javascript - python - java - c\# - php - android - html - jquery - c++ - css - ios - sql - mysql - r - reactjs - node.js - arrays - c - asp.net - json - python-3.x - .net - ruby-on-rails - sql-server - swift - django - angular - objective-c - excel - pandas - angularjs - regex - typescript - ruby - linux - ajax - iphone - vba - xml - laravel - spring - asp.net-mvc - database - wordpress - string - flutter - postgresql - mongodb - wpf - windows - xcode - amazon-web-services - bash - git - oracle-database - spring-boot - dataframe - azure - firebase - list - multithreading - docker - vb.net - react-native - eclipse - algorithm - powershell - macos - visual-studio - numpy - image - forms - scala - function - vue.js - performance - twitter-bootstrap - selenium - winforms - kotlin - loops - dart - express - hibernate - sqlite - matlab - python-2.7 - shell - rest - apache - entity-framework - android-studio - csv - maven - linq - qt - dictionary - unit-testing - asp.net-core - facebook - apache-spark - tensorflow - file - swing - class - unity-game-engine - sorting - date - authentication - go - symfony - t-sql - opencv - matplotlib - .htaccess - google-chrome - for-loop - datetime - codeigniter - perl - http - validation - sockets - google-maps - object - uitableview - xaml - oop - visual-studio-code - if-statement - cordova - ubuntu - web-services - email - android-layout - github - spring-mvc - elasticsearch - kubernetes - selenium-webdriver - ms-access - ggplot2 - user-interface - parsing - pointers - google-sheets - c++11 - security - machine-learning - google-apps-script - ruby-on-rails-3 - templates - flask - nginx - variables - exception - sql-server-2008 - gradle - debugging - tkinter - delphi - listview - jpa - asynchronous - haskell - web-scraping - pdf - jsp - ssl - amazon-s3 - google-cloud-platform - testing - jenkins - xamarin - wcf - batch-file - generics - npm - ionic-framework - network-programming - unix - recursion - google-app-engine - mongoose - visual-studio-2010 - .net-core - android-fragments - assembly - animation - math - svg - rust - session - intellij-idea - hadoop - join - curl - winapi - django-models - laravel-5 - next.js - url - heroku - http-redirect - tomcat - inheritance - google-cloud-firestore - webpack - gcc - image-processing - swiftui - keras - asp.net-mvc-4 - logging - dom - matrix - pyspark - actionscript-3 - button - post - optimization - firebase-realtime-database - jquery-ui - cocoa - xpath - iis - web - d3.js - javafx - firefox - xslt - internet-explorer - caching - select - asp.net-mvc-3 - opengl - events - asp.net-web-api - plot - dplyr - encryption - magento - stored-procedures - search - amazon-ec2 - ruby-on-rails-4 - memory - canvas - multidimensional-array - audio - random - jsf - vector - redux - cookies - input - facebook-graph-api - flash - indexing - xamarin.forms - arraylist - ipad - cocoa-touch - data-structures - video - model-view-controller - azure-devops - serialization - apache-kafka - jdbc - razor - woocommerce - awk - routes - mod-rewrite - servlets - excel-formula - beautifulsoup - filter - iframe - docker-compose - design-patterns - aws-lambda - text - visual-c++ - django-rest-framework - cakephp - mobile - android-intent - struct - react-hooks - methods - groovy - mvvm - lambda - ssh - checkbox - time - ecmascript-6 - grails - google-chrome-extension - installation - cmake - sharepoint - shiny - spring-security - jakarta-ee - plsql - android-recyclerview - core-data - types - sed - meteor - android-activity - bootstrap-4 - activerecord - websocket - replace - graph - group-by - scikit-learn - vim - file-upload - boost - junit - memory-management - sass - async-await - import - deep-learning - error-handling - eloquent - dynamic - dependency-injection - soap - silverlight - layout - apache-spark-sql - charts - deployment - browser - gridview - svn - while-loop - google-bigquery - vuejs2 - highcharts - dll - ffmpeg - view - foreach - makefile - plugins - redis - c\#-4.0 - reporting-services - jupyter-notebook - unicode - merge - reflection - https - server - google-maps-api-3 - twitter - oauth-2.0 - extjs - terminal - axios - pip - split - cmd - pytorch - encoding - django-views - collections - database-design - hash - netbeans - automation - data-binding - ember.js - build - tcp - pdo - sqlalchemy - apache-flex - entity-framework-core - concurrency - command-line - spring-data-jpa - printing - react-redux - java-8 - lua - html-table - neo4j - ansible - service - jestjs - mysqli - enums - parameters - flexbox - material-ui - module - promise - visual-studio-2012 - outlook - web-applications - firebase-authentication - uwp - webview - jquery-mobile - utf-8 - datatable - python-requests - parallel-processing - colors - drop-down-menu - scipy - scroll - tfs - hive - count - syntax - ms-word - twitter-bootstrap-3 - ssis - fonts - rxjs - constructor - google-analytics - file-io - paypal - three.js - powerbi - graphql - cassandra - discord - graphics - compiler-errors - gwt - react-router - socket.io - backbone.js - solr - memory-leaks - url-rewriting - datatables - nlp - terraform - oauth - datagridview - drupal - zend-framework - oracle11g - knockout.js - triggers - interface - neural-network - django-forms - casting - angular-material - jmeter - linked-list - google-api - path - timer - arduino - django-templates - orm - proxy - directory - windows-phone-7 - parse-platform - visual-studio-2015 - cron - conditional-statements - push-notification - functional-programming - primefaces - pagination - model - jar - xamarin.android - hyperlink - uiview - vbscript - visual-studio-2013 - google-cloud-functions - azure-active-directory - gitlab - jwt - download - swift3 - sql-server-2005 - process - configuration - rspec - pygame - properties - callback - combobox - windows-phone-8 - linux-kernel - safari - scrapy - permissions - emacs - x86 - clojure - scripting - raspberry-pi - scope - io - azure-functions - compilation - expo - responsive-design - mongodb-query - nhibernate - angularjs-directive - reference - request - bluetooth - binding - architecture - dns - playframework - 3d - pyqt - version-control - discord.js - doctrine-orm - package - f\# - rubygems - get - sql-server-2012 - autocomplete - tree - openssl - datepicker - kendo-ui - jackson - yii - controller - grep - nested - xamarin.ios - static - null - transactions - statistics - datagrid - active-directory - dockerfile - uiviewcontroller - webforms - sas - discord.py - computer-vision - phpmyadmin - notifications - duplicates - pycharm - mocking - youtube - yaml - nullpointerexception - menu - blazor - sum - plotly - bitmap - visual-studio-2008 - asp.net-mvc-5 - floating-point - css-selectors - yii2 - stl - electron - android-listview - jsf-2 - time-series - cryptography - ant - character-encoding - hashmap - stream - msbuild - asp.net-core-mvc - sdk - google-drive-api - jboss - selenium-chromedriver - joomla - devise - navigation - cors - cuda - anaconda - frontend - background - multiprocessing - binary - pyqt5 - camera - iterator - linq-to-sql - mariadb - onclick - android-jetpack-compose - ios7 - microsoft-graph-api - android-asynctask - rabbitmq - tabs - amazon-dynamodb - environment-variables - laravel-4 - uicollectionview - insert - linker - xsd - coldfusion - console - continuous-integration - upload - textview - ftp - opengl-es - operating-system - macros - mockito - localization - formatting - xml-parsing - json.net - type-conversion - vuejs3 - data.table - kivy - timestamp - integer - calendar - segmentation-fault - android-ndk - prolog - char - drag-and-drop - crash - jasmine - automated-tests - dependencies - geometry - azure-pipelines - fortran - android-gradle-plugin - itext - sprite-kit - mfc - header - attributes - nosql - firebase-cloud-messaging - format - nuxt.js - odoo - db2 - jquery-plugins - event-handling - julia - jenkins-pipeline - leaflet - annotations - flutter-layout - nestjs - keyboard - postman - arm - textbox - stripe-payments - visual-studio-2017 - gulp - libgdx - uikit - synchronization - timezone - azure-web-app-service - dom-events - wso2 - xampp - google-sheets-formula - crystal-reports - aggregation-framework - namespaces - swagger - android-emulator - uiscrollview - jvm - sequelize.js - chart.js - com - snowflake-cloud-data-platform - subprocess - geolocation - webdriver - garbage-collection - html5-canvas - sql-update - centos - dialog - concatenation - numbers - widget - qml - tuples - set - java-stream - mapreduce - ionic2 - smtp - windows-10 - android-edittext - nuget - rotation - spring-data - modal-dialog - radio-button - doctrine - http-headers - grid - lucene - sonarqube - xmlhttprequest - listbox - switch-statement - initialization - internationalization - components - boolean - apache-camel - google-play - gdb - serial-port - ios5 - ldap - return - youtube-api - pivot - eclipse-plugin - latex - frameworks - tags - containers - c++17 - subquery - github-actions - embedded - dataset - asp-classic - foreign-keys - label - uinavigationcontroller - delegates - copy - google-cloud-storage - struts2 - migration - protractor - base64 - uibutton - queue - find - sql-server-2008-r2 - arguments - composer-php - append - jaxb - stack - zip - tailwind-css - cucumber - autolayout - ide - entity-framework-6 - iteration - popup - r-markdown - windows-7 - vb6 - airflow - g++ - clang - hover - ssl-certificate - jqgrid - range - gmail Next You’ll be prompted to create an account to view your personalized homepage. 1. 1. [Home](https://stackoverflow.com/) 2. [Questions](https://stackoverflow.com/questions) 3. [AI Assist](https://stackoverflow.com/ai-assist) 4. [Tags](https://stackoverflow.com/tags) 5. [Challenges](https://stackoverflow.com/beta/challenges) 6. [Chat](https://chat.stackoverflow.com/?tab=explore) 7. [Articles](https://stackoverflow.blog/contributed?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=so-blog&utm_content=experiment-articles) 8. [Users](https://stackoverflow.com/users) 9. [Companies](https://stackoverflow.com/jobs/companies?so_medium=stackoverflow&so_source=SiteNav) 10. [Collectives]() 11. Communities for your favorite technologies. [Explore all Collectives](https://stackoverflow.com/collectives-all) 2. Stack Internal Stack Overflow for Teams is now called Stack Internal. Bring the best of human thought and AI automation together at your work. [Try for free](https://stackoverflowteams.com/teams/create/free/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=side-bar&utm_content=explore-teams) [Learn more](https://stackoverflow.co/internal/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=side-bar&utm_content=explore-teams) 3. [Stack Internal]() 4. Bring the best of human thought and AI automation together at your work. [Learn more](https://stackoverflow.co/internal/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=side-bar&utm_content=explore-teams-compact) ##### Collectives™ on Stack Overflow Find centralized, trusted content and collaborate around the technologies you use most. [Learn more about Collectives](https://stackoverflow.com/collectives) Stack Internal Knowledge at work Bring the best of human thought and AI automation together at your work. [Explore Stack Internal](https://stackoverflow.co/internal/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=side-bar&utm_content=explore-teams-compact-popover) # [How does one use Pytorch (+ cuda) with an A100 GPU?](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu) [Ask Question](https://stackoverflow.com/questions/ask) Asked 4 years, 9 months ago Modified [7 months ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu?lastactivity "2025-06-18 09:01:10Z") Viewed 75k times This question shows research effort; it is useful and clear 47 Save this question. Show activity on this post. I was trying to use my current code with an A100 gpu but I get this error: ``` Copy ``` which is reather confusing because it points to the usual pytorch installation but doesn't tell me which combination of pytorch version + cuda version to use for my specific hardware (A100). What is the right way to install pytorch for an A100? * These are some versions I've tried: ``` Copy ``` * note that this can be subtle because I've had this error with this machine + pytorch version in the past: [How to solve the famous \`unhandled cuda error, NCCL version 2.7.8\` error?](https://stackoverflow.com/questions/66807131/how-to-solve-the-famous-unhandled-cuda-error-nccl-version-2-7-8-error) * # Bonus 1: I still have errors: ``` Copy ``` one of the answers suggested to have nvcca & pytorch.version.cuda to match but they do not: ``` Copy ``` How do I match them? I this the error? Can someone display their pip, conda and nvcca version to see what set up works? More error messages: ``` Copy ``` after putting ``` Copy ``` - [python](https://stackoverflow.com/questions/tagged/python "show questions tagged 'python'") - [machine-learning](https://stackoverflow.com/questions/tagged/machine-learning "show questions tagged 'machine-learning'") - [neural-network](https://stackoverflow.com/questions/tagged/neural-network "show questions tagged 'neural-network'") - [pytorch](https://stackoverflow.com/questions/tagged/pytorch "show questions tagged 'pytorch'") [Share](https://stackoverflow.com/q/66992585 "Short permalink to this question") Share a link to this question Copy link [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/ "The current license for this post: CC BY-SA 4.0") [Improve this question](https://stackoverflow.com/posts/66992585/edit) Follow Follow this question to receive notifications [edited May 12, 2022 at 21:06](https://stackoverflow.com/posts/66992585/revisions "show all edits to this post") asked Apr 7, 2021 at 19:04 [![Charlie Parker's user avatar](https://i.sstatic.net/2B6rV.png?s=64)](https://stackoverflow.com/users/1601580/charlie-parker) [Charlie Parker](https://stackoverflow.com/users/1601580/charlie-parker) 6,2747979 gold badges264264 silver badges422422 bronze badges 3 - 2 Pytorch 1.7.0 or later with CUDA 11.0 or later [should work](https://www.marktechpost.com/2020/11/01/pytorch-releases-version-1-7-with-new-features-like-cuda-11-new-apis-for-ffts-and-nvidia-a100-generation-gpus-support/). Or you could use [NGC](https://ngc.nvidia.com/) Robert Crovella – [Robert Crovella](https://stackoverflow.com/users/1695960/robert-crovella "154,444 reputation") 2021-04-07 19:16:23 +00:00 [Commented Apr 7, 2021 at 19:16](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment118418045_66992585) - 1 @RobertCrovella if what you say it's true then the command needed is `conda install -y pytorch==1.7.1 torchvision torchaudio cudatoolkit=11.0 -c pytorch -c conda-forge` will try soon if it worked. Charlie Parker – [Charlie Parker](https://stackoverflow.com/users/1601580/charlie-parker "6,274 reputation") 2021-04-24 01:38:45 +00:00 [Commented Apr 24, 2021 at 1:38](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment118849441_66992585) - @CharlieParker just 14 minutes for the expiration of Bounty. non of these answers helpful? Sadra – [Sadra](https://stackoverflow.com/users/1076264/sadra "2,862 reputation") 2022-05-19 20:40:24 +00:00 [Commented May 19, 2022 at 20:40](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127748943_66992585) [Add a comment](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu "Use comments to ask for more information or suggest improvements. Avoid answering questions in comments.") \\| ## 6 Answers 6 Sorted by: [Reset to default](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu?answertab=scoredesc#tab-top) This answer is useful 48 Save this answer. Show activity on this post. From the link [pytorch site](https://pytorch.org/get-started/locally/) from @SimonB 's answer, I did: ``` Copypip3 install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 -f https://download.pytorch.org/whl/torch_stable.html ``` This solved the problem for me. [Share](https://stackoverflow.com/a/68626103 "Short permalink to this answer") Share a link to this answer Copy link [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/ "The current license for this post: CC BY-SA 4.0") [Improve this answer](https://stackoverflow.com/posts/68626103/edit) Follow Follow this answer to receive notifications answered Aug 2, 2021 at 18:02 [![James Hirschorn's user avatar](https://www.gravatar.com/avatar/e4ee321232e2b4700138d891f6d2f004?s=64&d=identicon&r=PG)](https://stackoverflow.com/users/1349673/james-hirschorn) [James Hirschorn](https://stackoverflow.com/users/1349673/james-hirschorn) 8,14466 gold badges5656 silver badges5858 bronze badges Sign up to request clarification or add additional context in comments. ## 6 Comments Add a comment [![](https://www.gravatar.com/avatar/fed3f0aad2fe6040187cd2a7c10e7dcf?s=48&d=identicon&r=PG&f=y&so-version=2)](https://stackoverflow.com/users/7998544/woma) Woma [Woma](https://stackoverflow.com/users/7998544/woma) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment121805600_68626103) For me, the conda installation did not work but the pip installation, no idea why 2021-08-25T11:32:10.157Z+00:00 1 Reply - Copy link [![](https://www.gravatar.com/avatar/aea0b2c6b6b44c5c7fe64f9a8561ffc8?s=48&d=identicon&r=PG&f=y&so-version=2)](https://stackoverflow.com/users/3188690/tian) Tian [Tian](https://stackoverflow.com/users/3188690/tian) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment125318618_68626103) Somehow, it does not work for now, seems that the download link not work for some reason: "returned a non-zero code: 137" 2022-01-28T01:53:45.39Z+00:00 0 Reply - Copy link [![](https://i.sstatic.net/2B6rV.png?s=64)](https://stackoverflow.com/users/1601580/charlie-parker) Charlie Parker [Charlie Parker](https://stackoverflow.com/users/1601580/charlie-parker) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127600966_68626103) are you sure this is right? I have that my pytorch wants 11.1 but nvcca is 11.0 see: `(meta_learning_a100) [miranda9@hal-dgx ~]$ python -c "import torch;print(torch.version.cuda)" 11.1 (meta_learning_a100) [miranda9@hal-dgx ~]$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2020 NVIDIA Corporation Built on Wed_Jul_22_19:09:09_PDT_2020 Cuda compilation tools, release 11.0, V11.0.221 Build cuda_11.0_bu.TC445_37.28845127_0` 2022-05-12T20:43:29.293Z+00:00 0 Reply - Copy link [![](https://www.gravatar.com/avatar/e4ee321232e2b4700138d891f6d2f004?s=48&d=identicon&r=PG)](https://stackoverflow.com/users/1349673/james-hirschorn) James Hirschorn [James Hirschorn](https://stackoverflow.com/users/1349673/james-hirschorn) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127668237_68626103) @CharlieParker It's been too long for me to recall the context for this question. But `nvidia-smi` reveals that I have CUDA 11.4 and nvcc 10.1 2022-05-16T16:26:26.62Z+00:00 0 Reply - Copy link [![](https://www.gravatar.com/avatar/3b495f8e4707e59490dda3cd0940c5e9?s=48&d=identicon&r=PG&f=y&so-version=2)](https://stackoverflow.com/users/8183621/pascaliv) PascalIv [PascalIv](https://stackoverflow.com/users/8183621/pascaliv) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment134468011_68626103) ERROR: No matching distribution found for torch==1.9.0+cu111 2023-05-15T09:12:04.423Z+00:00 2 Reply - Copy link Add a comment \\| Show 1 more comment This answer is useful 8 Save this answer. Show activity on this post. I've got an A100 and have had success with ``` Copyconda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c nvidia ``` Which is now also recommended on the [pytorch site](https://pytorch.org/get-started/locally/) [Share](https://stackoverflow.com/a/67521477 "Short permalink to this answer") Share a link to this answer Copy link [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/ "The current license for this post: CC BY-SA 4.0") [Improve this answer](https://stackoverflow.com/posts/67521477/edit) Follow Follow this answer to receive notifications answered May 13, 2021 at 14:58 [![Simon B's user avatar](https://www.gravatar.com/avatar/4a035074530513cee4b47303f64f904a?s=64&d=identicon&r=PG)](https://stackoverflow.com/users/1264592/simon-b) [Simon B](https://stackoverflow.com/users/1264592/simon-b) 26711 silver badge77 bronze badges ## 1 Comment Add a comment [![](https://i.sstatic.net/2B6rV.png?s=64)](https://stackoverflow.com/users/1601580/charlie-parker) Charlie Parker [Charlie Parker](https://stackoverflow.com/users/1601580/charlie-parker) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127600972_67521477) are you sure this is right? I have that my pytorch wants 11.1 but nvcca is 11.0 see: `(meta_learning_a100) [miranda9@hal-dgx ~]$ python -c "import torch;print(torch.version.cuda)" 11.1 (meta_learning_a100) [miranda9@hal-dgx ~]$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2020 NVIDIA Corporation Built on Wed_Jul_22_19:09:09_PDT_2020 Cuda compilation tools, release 11.0, V11.0.221 Build cuda_11.0_bu.TC445_37.28845127_0` 2022-05-12T20:43:54.453Z+00:00 0 Reply - Copy link This answer is useful 4 Save this answer. Show activity on this post. To me this is what worked: ``` Copy ``` then I tested it, asked for the device and did a matrix multiply, no errors is it worked: ``` Copy ``` gpu pytorch code: ``` Copy ``` [Share](https://stackoverflow.com/a/70946283 "Short permalink to this answer") Share a link to this answer Copy link [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/ "The current license for this post: CC BY-SA 4.0") [Improve this answer](https://stackoverflow.com/posts/70946283/edit) Follow Follow this answer to receive notifications answered Feb 1, 2022 at 19:18 [![Charlie Parker's user avatar](https://i.sstatic.net/2B6rV.png?s=64)](https://stackoverflow.com/users/1601580/charlie-parker) [Charlie Parker](https://stackoverflow.com/users/1601580/charlie-parker) 6,2747979 gold badges264264 silver badges422422 bronze badges ## 1 Comment Add a comment [![](https://lh4.googleusercontent.com/-6pGlNM6cH_4/AAAAAAAAAAI/AAAAAAAAACU/YixjVAUWv20/s48-rj/photo.jpg)](https://stackoverflow.com/users/9058482/harsh-kumar-chourasia) Harsh Kumar Chourasia [Harsh Kumar Chourasia](https://stackoverflow.com/users/9058482/harsh-kumar-chourasia) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment129644183_70946283) Thanks! This solution worked for me. I was trying to run pytorch inside docker. I uninstalled all the cuda libraries and pre-installed torch before running this command. 2022-08-19T05:48:05.663Z+00:00 0 Reply - Copy link This answer is useful 2 Save this answer. Show activity on this post. I had the same problem. You need to install CUDA 11.0 instead of 10.2 and reinstall PyTorch for this CUDA version. [Share](https://stackoverflow.com/a/67026338 "Short permalink to this answer") Share a link to this answer Copy link [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/ "The current license for this post: CC BY-SA 4.0") [Improve this answer](https://stackoverflow.com/posts/67026338/edit) Follow Follow this answer to receive notifications answered Apr 9, 2021 at 18:15 [![guillaumefrd's user avatar](https://lh3.googleusercontent.com/-Q2BzoBitSPg/AAAAAAAAAAI/AAAAAAAAX1U/mQloaeFMW24/s64-rj/photo.jpg)](https://stackoverflow.com/users/8926322/guillaumefrd) [guillaumefrd](https://stackoverflow.com/users/8926322/guillaumefrd) 12766 bronze badges ## 3 Comments Add a comment [![](https://i.sstatic.net/2B6rV.png?s=64)](https://stackoverflow.com/users/1601580/charlie-parker) Charlie Parker [Charlie Parker](https://stackoverflow.com/users/1601580/charlie-parker) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment118848956_67026338) did you install pytorch 1.8.0 using cuda 11.0 or pytorch 1.7.x? 2021-04-24T00:40:10.057Z+00:00 0 Reply - Copy link [![](https://lh3.googleusercontent.com/-Q2BzoBitSPg/AAAAAAAAAAI/AAAAAAAAX1U/mQloaeFMW24/s48-rj/photo.jpg)](https://stackoverflow.com/users/8926322/guillaumefrd) guillaumefrd [guillaumefrd](https://stackoverflow.com/users/8926322/guillaumefrd) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment118869529_67026338) I tried 1.8.0 and 1.7.1, both were working. 2021-04-25T06:40:29.923Z+00:00 1 Reply - Copy link [![](https://i.sstatic.net/2B6rV.png?s=64)](https://stackoverflow.com/users/1601580/charlie-parker) Charlie Parker [Charlie Parker](https://stackoverflow.com/users/1601580/charlie-parker) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127600974_67026338) are you sure this is right? I have that my pytorch wants 11.1 but nvcca is 11.0 see: `(meta_learning_a100) [miranda9@hal-dgx ~]$ python -c "import torch;print(torch.version.cuda)" 11.1 (meta_learning_a100) [miranda9@hal-dgx ~]$ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2020 NVIDIA Corporation Built on Wed_Jul_22_19:09:09_PDT_2020 Cuda compilation tools, release 11.0, V11.0.221 Build cuda_11.0_bu.TC445_37.28845127_0` 2022-05-12T20:44:09.713Z+00:00 0 Reply - Copy link Add a comment This answer is useful 2 Save this answer. Show activity on this post. This solution is tested on a multi GPU A100** environment: create a clean conda environment: `conda create -n pya100 python=3.9` then check your nvcc version by: `nvcc --version` \#mine return `11.3` then install pytorch in this way: (as of now it installs Pytorch 1.11.0, torchvision 0.12.0) ``` Copyconda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch -c nvidia ``` now `python -c "import torch;print(torch.version.cuda)"` returns `11.3` (though I don't think it matters that much) I shared my environment file [Here](https://pastebin.com/qTRzFHjZ). You can build one environment based on using this: (just replace NAMEOFENVIRONMENT with your environment name) ``` Copyconda env update --name NAMEOFENVIRONMENT --file environment.yml ``` [Share](https://stackoverflow.com/a/72244310 "Short permalink to this answer") Share a link to this answer Copy link [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/ "The current license for this post: CC BY-SA 4.0") [Improve this answer](https://stackoverflow.com/posts/72244310/edit) Follow Follow this answer to receive notifications [edited May 18, 2022 at 2:24](https://stackoverflow.com/posts/72244310/revisions "show all edits to this post") answered May 14, 2022 at 22:34 [![Sadra's user avatar](https://i.sstatic.net/M7tOf.jpg?s=64)](https://stackoverflow.com/users/1076264/sadra) [Sadra](https://stackoverflow.com/users/1076264/sadra) 2,86222 gold badges2525 silver badges3434 bronze badges ## 3 Comments Add a comment [![](https://i.sstatic.net/M7tOf.jpg?s=64)](https://stackoverflow.com/users/1076264/sadra) Sadra [Sadra](https://stackoverflow.com/users/1076264/sadra) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127638212_72244310) @CharlieParker I think this address your problem 2022-05-14T22:35:27.917Z+00:00 0 Reply - Copy link [![](https://i.sstatic.net/M7tOf.jpg?s=64)](https://stackoverflow.com/users/1076264/sadra) Sadra [Sadra](https://stackoverflow.com/users/1076264/sadra) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127648326_72244310) if the current solution did not work please share `apt list --installed` through pastebin so I can compare if mine and give it another try 2022-05-15T17:20:42.75Z+00:00 0 Reply - Copy link [![](https://i.sstatic.net/M7tOf.jpg?s=64)](https://stackoverflow.com/users/1076264/sadra) Sadra [Sadra](https://stackoverflow.com/users/1076264/sadra) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127748922_72244310) @CharlieParker does it help? 2022-05-19T20:39:31.213Z+00:00 0 Reply - Copy link Add a comment This answer is useful 1 Save this answer. Show activity on this post. Check your installed version of torch, torchvision, torchaudio etc. using ``` Copy<your virtualenv path>/bin/python -m torch.utils.collect_env ``` In my case I had this - ``` Copy ``` Since, I was not using torchvision or torchaudio, I just updated my `torch` version using the suggestion by @JamesHirschorn and selected the one according to my `torch` version from this [pytorch](https://download.pytorch.org/whl/torch_stable.html) link. e.g. in my case, the `torch` version was 1.11.0 and hence I installed `torch==1.11.0+cu113` `pip install torch==1.11.0+cu113 -f https://download.pytorch.org/whl/torch_stable.html` After update, the output of `<your virtualenv path>/bin/python -m torch.utils.collect_env` was ``` Copy ``` [Share](https://stackoverflow.com/a/72106714 "Short permalink to this answer") Share a link to this answer Copy link [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/ "The current license for this post: CC BY-SA 4.0") [Improve this answer](https://stackoverflow.com/posts/72106714/edit) Follow Follow this answer to receive notifications answered May 4, 2022 at 0:21 [![rhn89's user avatar](https://i.sstatic.net/HDWUs.jpg?s=64)](https://stackoverflow.com/users/3082461/rhn89) [rhn89](https://stackoverflow.com/users/3082461/rhn89) 44855 silver badges1111 bronze badges ## 1 Comment Add a comment [![](https://i.sstatic.net/6rTV9.png?s=64)](https://stackoverflow.com/users/979474/user956584) user956584 [user956584](https://stackoverflow.com/users/979474/user956584) [Over a year ago](https://stackoverflow.com/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu#comment127424655_72106714) for me with message: Tesla K20Xm with CUDA capability sm\_35 is not compatible with the current PyTorch installation. this works: pip install torch==1.2.0+cu92 -f [download.pytorch.org/whl/torch\_stable.html](https://download.pytorch.org/whl/torch_stable.html) 2022-05-04T18:03:05.943Z+00:00 1 Reply - Copy link Start asking to get answers Find the answer to your question by asking. [Ask question](https://stackoverflow.com/questions/ask) Explore related questions - [python](https://stackoverflow.com/questions/tagged/python "show questions tagged 'python'") - [machine-learning](https://stackoverflow.com/questions/tagged/machine-learning "show questions tagged 'machine-learning'") - [neural-network](https://stackoverflow.com/questions/tagged/neural-network "show questions tagged 'neural-network'") - [pytorch](https://stackoverflow.com/questions/tagged/pytorch "show questions tagged 'pytorch'") See similar questions with these tags. - The Overflow Blog - [Is that allowed? Authentication and authorization in Model Context Protocol](https://stackoverflow.blog/2026/01/21/is-that-allowed-authentication-and-authorization-in-model-context-protocol/?cb=1) - [Automate your security whack-a-mole: Q\&A with Exaforce](https://stackoverflow.blog/2026/01/22/automate-your-security-whack-a-mole-q-and-a-with-exaforce/?cb=1) - Featured on Meta - [AI Assist: recent updates (January 2026)](https://meta.stackexchange.com/questions/416954/ai-assist-recent-updates-january-2026?cb=1) - [Community Engagement Across the Network: Focus for 2026](https://meta.stackexchange.com/questions/416985/community-engagement-across-the-network-focus-for-2026?cb=1) - [All users on Stack Overflow can now participate in chat](https://meta.stackoverflow.com/questions/437932/all-users-on-stack-overflow-can-now-participate-in-chat?cb=1) - [Policy: Generative AI (e.g., ChatGPT) is banned](https://meta.stackoverflow.com/questions/421831/policy-generative-ai-e-g-chatgpt-is-banned?cb=1) - [Stack Overflow now uses machine learning to flag spam automatically](https://meta.stackoverflow.com/questions/437959/stack-overflow-now-uses-machine-learning-to-flag-spam-automatically?cb=1) - [No, I do not believe this is the end](https://meta.stackoverflow.com/questions/438009/no-i-do-not-believe-this-is-the-end?cb=1) Community activity Last 1 hr - Users online activity 5257 users online - 11 questions - 16 answers - 26 comments - 156 upvotes Popular tags [image-processing](https://stackoverflow.com/questions/tagged/image-processing)[excel](https://stackoverflow.com/questions/tagged/excel)[png](https://stackoverflow.com/questions/tagged/png)[html](https://stackoverflow.com/questions/tagged/html)[client-side](https://stackoverflow.com/questions/tagged/client-side)[python](https://stackoverflow.com/questions/tagged/python) Popular unanswered question [MongoDB querySrv ECONNREFUSED on Node.js 22 after switching Node versions (worked before, now fails)](https://stackoverflow.com/questions/79873598) [node.js](https://stackoverflow.com/questions/tagged/node.js)[mongodb](https://stackoverflow.com/questions/tagged/mongodb)[windows](https://stackoverflow.com/questions/tagged/windows)[express](https://stackoverflow.com/questions/tagged/express)[mongoose](https://stackoverflow.com/questions/tagged/mongoose) [![User avatar](https://www.gravatar.com/avatar/66dea90129372b0e057d203049c3142b?s=256&d=identicon&r=PG&f=y&so-version=2)](https://stackoverflow.com/users/32237460) [Vin](https://stackoverflow.com/users/32237460) - 11 19 hours ago #### Linked [23](https://stackoverflow.com/questions/66807131/how-to-solve-the-famous-unhandled-cuda-error-nccl-version-2-7-8-error?lq=1 "Question score (upvotes - downvotes)") [How to solve the famous \`unhandled cuda error, NCCL version 2.7.8\` error?](https://stackoverflow.com/questions/66807131/how-to-solve-the-famous-unhandled-cuda-error-nccl-version-2-7-8-error?noredirect=1&lq=1) [0](https://stackoverflow.com/questions/76366411/install-pytorch-dependencies-torchtext-torchdata-torch-with-cuda-and-a100?lq=1 "Question score (upvotes - downvotes)") [Install Pytorch dependencies (torchtext + torchdata + torch) with cuda and A100 GPU](https://stackoverflow.com/questions/76366411/install-pytorch-dependencies-torchtext-torchdata-torch-with-cuda-and-a100?noredirect=1&lq=1) #### Related [1](https://stackoverflow.com/questions/50771001/pytorch-not-using-cuda-device?rq=3 "Question score (upvotes - downvotes)") [Pytorch not using cuda device](https://stackoverflow.com/questions/50771001/pytorch-not-using-cuda-device?rq=3) [46](https://stackoverflow.com/questions/50954479/using-cuda-with-pytorch?rq=3 "Question score (upvotes - downvotes)") [Using CUDA with pytorch?](https://stackoverflow.com/questions/50954479/using-cuda-with-pytorch?rq=3) [12](https://stackoverflow.com/questions/51605893/why-doesnt-my-simple-pytorch-network-work-on-gpu-device?rq=3 "Question score (upvotes - downvotes)") [Why doesn't my simple pytorch network work on GPU device?](https://stackoverflow.com/questions/51605893/why-doesnt-my-simple-pytorch-network-work-on-gpu-device?rq=3) [4](https://stackoverflow.com/questions/56852347/model-cuda-in-pytorch?rq=3 "Question score (upvotes - downvotes)") [model.cuda() in pytorch](https://stackoverflow.com/questions/56852347/model-cuda-in-pytorch?rq=3) [2](https://stackoverflow.com/questions/59497887/how-can-i-convert-pytorch-cpu-based-transformation-to-cuda-based?rq=3 "Question score (upvotes - downvotes)") [How can I convert pytorch cpu-based transformation to cuda-based?](https://stackoverflow.com/questions/59497887/how-can-i-convert-pytorch-cpu-based-transformation-to-cuda-based?rq=3) [0](https://stackoverflow.com/questions/60063910/forcing-pytorch-to-use-gpu?rq=3 "Question score (upvotes - downvotes)") [Forcing Pytorch to use GPU](https://stackoverflow.com/questions/60063910/forcing-pytorch-to-use-gpu?rq=3) [0](https://stackoverflow.com/questions/62476250/how-to-run-a-pre-trained-pytorch-model-on-the-gpu?rq=3 "Question score (upvotes - downvotes)") [How to run a pre-trained pytorch model on the GPU?](https://stackoverflow.com/questions/62476250/how-to-run-a-pre-trained-pytorch-model-on-the-gpu?rq=3) [25](https://stackoverflow.com/questions/63423463/using-pytorch-cuda-on-macbook-pro?rq=3 "Question score (upvotes - downvotes)") [Using pytorch Cuda on MacBook Pro](https://stackoverflow.com/questions/63423463/using-pytorch-cuda-on-macbook-pro?rq=3) [4](https://stackoverflow.com/questions/67645531/how-does-one-run-pytorch-on-a-a40-gpu-without-errors-with-ddp-too?rq=3 "Question score (upvotes - downvotes)") [How does one run PyTorch on a A40 GPU without errors (with DDP too)?](https://stackoverflow.com/questions/67645531/how-does-one-run-pytorch-on-a-a40-gpu-without-errors-with-ddp-too?rq=3) [0](https://stackoverflow.com/questions/69591391/pytorch-using-cuda-prevents-optimization-from-working?rq=3 "Question score (upvotes - downvotes)") [Pytorch: using CUDA prevents optimization from working](https://stackoverflow.com/questions/69591391/pytorch-using-cuda-prevents-optimization-from-working?rq=3) #### [Hot Network Questions](https://stackexchange.com/questions?tab=hot) - [Question about Einstein problem](https://mathoverflow.net/questions/507294/question-about-einstein-problem) - [Is there a deeper reason why surjections are harder to count than injections?](https://math.stackexchange.com/questions/5120661/is-there-a-deeper-reason-why-surjections-are-harder-to-count-than-injections) - [Abandoned art and copyright?](https://law.stackexchange.com/questions/113979/abandoned-art-and-copyright) - [Have there been major developments in the Langlands Program in the number field setting since 2010?](https://mathoverflow.net/questions/507242/have-there-been-major-developments-in-the-langlands-program-in-the-number-field) - [JSON Parser in C++](https://codereview.stackexchange.com/questions/301080/json-parser-in-c) - [Workbench on Spring'26 sandbox: OAUTH\_APPROVAL\_ERROR\_GENERIC](https://salesforce.stackexchange.com/questions/438640/workbench-on-spring26-sandbox-oauth-approval-error-generic) - [Is there any way of moving the symbol vertically wrt the notehead?](https://music.stackexchange.com/questions/143091/is-there-any-way-of-moving-the-symbol-vertically-wrt-the-notehead) - [How do I stop the radidons beak from cracking?](https://worldbuilding.stackexchange.com/questions/272416/how-do-i-stop-the-radidons-beak-from-cracking) - [How can I maintain tension in a PF2e, a system with unlimited out-of-combat healing?](https://rpg.stackexchange.com/questions/218576/how-can-i-maintain-tension-in-a-pf2e-a-system-with-unlimited-out-of-combat-heal) - [Dual passports (UK/HK) holder entering & leaving the US](https://travel.stackexchange.com/questions/203124/dual-passports-uk-hk-holder-entering-leaving-the-us) - ["Sed nescio quomodo nihil tam absurde dici potest quod non dicatur ab aliquo philosophorum." - Why "quod" and not "ut" or "quam"?](https://latin.stackexchange.com/questions/27062/sed-nescio-quomodo-nihil-tam-absurde-dici-potest-quod-non-dicatur-ab-aliquo-phi) - [(I Cor 2:16) What actually \is\ the "mind of Christ"? Is it the Spirit of knowledge, wisdom, understanding, counsel and might in Isa 11:2?](https://hermeneutics.stackexchange.com/questions/114138/i-cor-216-what-actually-is-the-mind-of-christ-is-it-the-spirit-of-knowle) - [Looking for an old physics lab meme (early 2000s)](https://physics.stackexchange.com/questions/868126/looking-for-an-old-physics-lab-meme-early-2000s) - [Batcher's sort in Java: a θ(N log N log N) sorting algorithm](https://codereview.stackexchange.com/questions/301081/batchers-sort-in-java-a-%CE%B8n-log-n-log-n-sorting-algorithm) - [Arrange 11 (or fewer) squares that will be colored](https://puzzling.stackexchange.com/questions/136808/arrange-11-or-fewer-squares-that-will-be-colored) - [What did John Austin (died 1859) mean by habit, habitual(ly)?](https://philosophy.stackexchange.com/questions/135453/what-did-john-austin-died-1859-mean-by-habit-habitually) - [Coming-of-age short story involving a boy abducted by aliens](https://scifi.stackexchange.com/questions/302964/coming-of-age-short-story-involving-a-boy-abducted-by-aliens) - [Approximating a polynomial with one with a particular Galois group](https://mathoverflow.net/questions/507277/approximating-a-polynomial-with-one-with-a-particular-galois-group) - [GCC miscompiles long chains of overloaded \`operator,\`](https://stackoverflow.com/questions/79872387/gcc-miscompiles-long-chains-of-overloaded-operator) - [Efficiency when converting camelCase to snake\_case](https://codereview.stackexchange.com/questions/301067/efficiency-when-converting-camelcase-to-snake-case) - [My direct manager accepts my timesheet, RM goes back in and takes away some of my hours](https://workplace.stackexchange.com/questions/203103/my-direct-manager-accepts-my-timesheet-rm-goes-back-in-and-takes-away-some-of-m) - [Where are the words? A connect wall](https://puzzling.stackexchange.com/questions/136800/where-are-the-words-a-connect-wall) - [Intuitive analysis of d'Alembert's solution to the 1+1D wave equation](https://physics.stackexchange.com/questions/868193/intuitive-analysis-of-dalemberts-solution-to-the-11d-wave-equation) - [Construction of a large stealth spacecraft](https://worldbuilding.stackexchange.com/questions/272415/construction-of-a-large-stealth-spacecraft) [Question feed](https://stackoverflow.com/feeds/question/66992585 "Feed of this question and its answers") # Subscribe to RSS Question feed To subscribe to this RSS feed, copy and paste this URL into your RSS reader. ![](https://stackoverflow.com/posts/66992585/ivc/b2a4?prg=f420b4f9-cf6a-45f5-a265-d6a1b7a03d68) lang-py # Why are you flagging this comment? Probable spam. This comment promotes a product, service or website while [failing to disclose the author's affiliation](https://stackoverflow.com/help/promotion). Unfriendly or contains harassment/bigotry/abuse. This comment is unkind, insulting or attacks another person or group. Learn more in our [Abusive behavior policy](https://stackoverflow.com/conduct/abusive-behavior). Not needed. This comment is not relevant to the post. ``` ``` Enter at least 6 characters Something else. A problem not listed above. Try to be as specific as possible. ``` ``` Enter at least 6 characters Flag comment Cancel You have 0 flags left today ![](https://stackoverflow.com/js-false.gif) ![](https://stackoverflow.com/js-true.gif) # ![Illustration of upvote icon after it is clicked](https://stackoverflow.com/Content/Img/modal/img-upvote.png?v=fce73bd9724d) # Hang on, you can't upvote just yet. You'll need to complete a few actions and gain 15 reputation points before being able to upvote. Upvoting indicates when questions and answers are useful. [What's reputation and how do I get it?](https://stackoverflow.com/help/whats-reputation) Instead, you can save this post to reference later. Save this post for later Not now ##### [Stack Overflow](https://stackoverflow.com/) - [Questions](https://stackoverflow.com/questions) - [Help](https://stackoverflow.com/help) - [Chat](https://chat.stackoverflow.com/?tab=explore) ##### [Business](https://stackoverflow.co/) - [Stack Internal](https://stackoverflow.co/internal/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=footer&utm_content=teams) - [Stack Data Licensing](https://stackoverflow.co/data-licensing/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=footer&utm_content=data-licensing) - [Stack Ads](https://stackoverflow.co/advertising/?utm_medium=referral&utm_source=stackoverflow-community&utm_campaign=footer&utm_content=advertising) ##### [Company](https://stackoverflow.co/) - [About](https://stackoverflow.co/) - [Press](https://stackoverflow.co/company/press/) - [Work Here](https://stackoverflow.co/company/work-here/) - [Legal](https://stackoverflow.com/legal) - [Privacy Policy](https://stackoverflow.com/legal/privacy-policy) - [Terms of Service](https://stackoverflow.com/legal/terms-of-service/public) - [Contact Us](https://stackoverflow.com/contact) - Cookie Settings - [Cookie Policy](https://policies.stackoverflow.co/stack-overflow/cookie-policy) ##### [Stack Exchange Network](https://stackexchange.com/) - [Technology](https://stackexchange.com/sites#technology) - [Culture & recreation](https://stackexchange.com/sites#culturerecreation) - [Life & arts](https://stackexchange.com/sites#lifearts) - [Science](https://stackexchange.com/sites#science) - [Professional](https://stackexchange.com/sites#professional) - [Business](https://stackexchange.com/sites#business) - [API](https://api.stackexchange.com/) - [Data](https://data.stackexchange.com/) - [Blog](https://stackoverflow.blog/?blb=1) - [Facebook](https://www.facebook.com/officialstackoverflow/) - [Twitter](https://twitter.com/stackoverflow) - [LinkedIn](https://linkedin.com/company/stack-overflow) - [Instagram](https://www.instagram.com/thestackoverflow) Site design / logo © 2026 Stack Exchange Inc; user contributions licensed under [CC BY-SA](https://stackoverflow.com/help/licensing) . rev 2026.1.22.38918
Readable Markdown	null
Shard	169 (laksa)
Root Hash	714406497480128969
Unparsed URL	com,stackoverflow!/questions/66992585/how-does-one-use-pytorch-cuda-with-an-a100-gpu s443