Forked from garg-aayush/Steps_multiple_cuda_environments.md
Created
September 26, 2024 06:14
Revisions
-
garg-aayush revised this gist
May 19, 2024 . No changes.There are no files selected for viewing
-
May 19, 2024 . 1 changed file with 87 additions and 48 deletions.There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode charactersOriginal file line number Diff line number Diff line change @@ -1,7 +1,9 @@ # Steps to manage multiple CUDA environments > Latest Update: May 19th, 2024 This gist contains all the steps required to: - Install multiple CUDA versions (e.g., `CUDA 11.8 and `CUDA 12.1 - Manage multiple CUDA environments on Ubuntu using the utility called [environment modules](https://modules.readthedocs.io/en/latest/). - Use this approach to avoid CUDA environment conflicts. @@ -16,97 +18,104 @@ This gist contains all the steps required to: - Check GPU and available drives ```bash ubuntu-drivers devices # install it using: sudo ubuntu-drivers ``` - Install the compatible driver ```bash # best to allow Ubuntu to autodetect and install the compatible nvidia-driver sudo ubuntu-drivers install ``` > For example, I tried to install `nvidia-driver-545` using `sudo ubuntu-drivers install nvidia:545` command. However, I was unable to install it. There was always some or the other issue. > **Note**: > Please **restart** your system after installing the nvidia driver. Ideally, you should be able to get GPU state and stats using `nvidia-smi` - Check the installed NVIDIA driver ```bash nvidia-detector ``` > - Additionally, you can also install NVIDIA drivers using the **Software & Updates** Ubuntu app. Just go to the **Additional Drivers** tab, choose a driver, and click **Apply Changes**. ## 2. Install `CUDA 11.8` and `CUDA 12.1` - Go to the [https://developer.nvidia.com/cuda-toolkit-archive](https://developer.nvidia.com/cuda-toolkit-archive) and select `CUDA Toolkit 11.8` from the available options. - Choose your OS, architecture, distribution, version, and installer type. For example, in my case: Option | value | :---:|:---:| | OS | Linux | | Architecture | x86_64 | | Distribution | Linux | | Version | 22.04 | | Installer type | deb(local) | - Follow the provided installation instructions by copying and pasting the commands into your terminal. This will install `CUDA 11.8`. Use the following commands: ```bash wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-ubuntu2204.pin sudo mv cuda-ubuntu2204.pin /etc/apt/preferences.d/cuda-repository-pin-600 wget https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda-repo-ubuntu2204-11-8-local_11.8.0-520.61.05-1_amd64.deb sudo dpkg -i cuda-repo-ubuntu2204-11-8-local_11.8.0-520.61.05-1_amd64.deb sudo cp /var/cuda-repo-ubuntu2204-11-8-local/cuda-*-keyring.gpg /usr/share/keyrings/ sudo apt-get update sudo apt-get -y install cuda ``` - Similarly, install `CUDA 12.1` using the following commands: ```bash wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-ubuntu2204.pin sudo mv cuda-ubuntu2204.pin /etc/apt/preferences.d/cuda-repository-pin-600 wget https://developer.download.nvidia.com/compute/cuda/12.1.0/local_installers/cuda-repo-ubuntu2204-12-1-local_12.1.0-530.30.02-1_amd64.deb sudo dpkg -i cuda-repo-ubuntu2204-12-1-local_12.1.0-530.30.02-1_amd64.deb sudo cp /var/cuda-repo-ubuntu2204-12-1-local/cuda-*-keyring.gpg /usr/share/keyrings/ sudo apt-get update sudo apt-get -y install cuda ``` - Make sure to copy and execute the commands above in your terminal to install `CUDA 11.8` and `CUDA 12.1` on your system. ## 3. Install `cuDNN` library - Go to https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/linux-x86_64/ and download the `cuDNN` tar for `CUDA 11.x`. Note that you might need to create a developer's account first. - Untar the downloaded file using the following command: ```bash tar -xvf cudnn-linux-x86_64-9.1.0.70_cuda11-archive.tar.xz # CUDA 11.x tar -xvf cudnn-linux-x86_64-9.1.0.70_cuda12-archive.tar.xz # CUDA 12.x ``` - Copy the `cuDNN` files to the `CUDA` toolkit files: ```bash # for CUDA 11.8 cd cudnn-linux-x86_64-9.1.0.70_cuda11-archive/ sudo cp include/cudnn*.h /usr/local/cuda-11.8/include sudo cp lib64/libcudnn* /usr/local/cuda-11.8/lib64 # for CUDA 12.1 cd cudnn-linux-x86_64-9.1.0.70_cuda12-archive/ sudo cp include/cudnn*.h /usr/local/cuda-12.1/include sudo cp lib64/libcudnn* /usr/local/cuda-12.1/lib64 ``` - Make the files executable: ```bash sudo chmod a+r /usr/local/cuda-11.8/include/cudnn*.h /usr/local/cuda-11.8/lib64/libcudnn* sudo chmod a+r /usr/local/cuda-12.1/include/cudnn*.h /usr/local/cuda-12.1/lib64/libcudnn* ``` > **Note**: > Strictly speaking, you are done with the CUDA setup. You can use it by adding the CUDA bin and library path to the PATH and LD_LIBRARY_PATH environment variables. For example, you can set up CUDA 11.8 by adding the following lines in the `~/.bashrc`: > ```bash > PATH=/usr/local/cuda-11.8/bin:$PATH > LD_LIBRARY_PATH=/usr/local/cuda-11.8/extras/CUPTI/lib64:$LD_LIBRARY_PATH > LD_LIBRARY_PATH=/usr/local/cuda-11.8/lib64:$LD_LIBRARY_PATH > ``` > Similarly, you can set up CUDA 12.1. However, manually changing the paths every time can be cumbersome! > **Note**: In case, you only want to install either of the one, CUDNN 11.x or CUDNN 12.x. The simpler way is to go to https://developer.nvidia.com/cudnn-downloads and install the CUDNN 11.x or CUDNN 12.x similar to CUDA installation. ## 4. Manage multiple CUDA versions using `environment modules` ### a) Install the environment modules utility: - Run the following commands: ```bash @@ -130,11 +139,11 @@ This gist contains all the steps required to: sudo mkdir -p /usr/share/modules/modulefiles/cuda ``` - Create a modulefile `/usr/share/modules/modulefiles/cuda/11.8` for `CUDA 11.8` and add the following lines: ```bash #%Module1.0 ## ## cuda 11.8 modulefile ## proc ModulesHelp { } { @@ -145,12 +154,12 @@ This gist contains all the steps required to: module-whatis "sets up environment for CUDA 11.8" if { [ is-loaded cuda/12.1 ] } { module unload cuda/12.1 } set version 11.8 set root /usr/local/cuda-11.8 setenv CUDA_HOME $root prepend-path PATH $root/bin @@ -159,11 +168,11 @@ This gist contains all the steps required to: conflict cuda ``` - Similarly, create a modulefile `/usr/share/modules/modulefiles/cuda/12.1` for `CUDA 12.1` and add the following lines: ```bash #%Module1.0 ## ## cuda 12.1 modulefile ## proc ModulesHelp { } { @@ -172,14 +181,14 @@ This gist contains all the steps required to: puts stderr "\tSets up environment for CUDA $version\n" } module-whatis "sets up environment for CUDA 12.1" if { [ is-loaded cuda/11.8 ] } { module unload cuda/11.8 } set version 12.1 set root /usr/local/cuda-12.1 setenv CUDA_HOME $root prepend-path PATH $root/bin @@ -196,6 +205,7 @@ This gist contains all the steps required to: ``` > **Note**: make sure to reload your terminal. ## 5. Changing and Viewing the CUDA Module - To change and view the loaded CUDA module, you can use the following commands: ```bash @@ -205,7 +215,7 @@ This gist contains all the steps required to: module avail # Load a specific cuda version module load cuda/12.1 # Unload the currently loaded CUDA module module unload cuda # Load CUDA 11.8 @@ -217,4 +227,33 @@ This gist contains all the steps required to: echo $PATH echo $LD_LIBRARY_PATH ``` > **Note**: You can add additional `CUDA` versions or other packages by creating corresponding modulefiles and following the steps outlined in this gist. ## 6. Some Useful Tips ### a) `nvidia-smi` does not works - Sometime, after Ubuntu update or some other weird issue. The system might not be able to detect drivers. For example, you get erros such as `nvidia-smi has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.` The best solution is to remove the current drivers and reinstall the compatible nvidia-driver. ```bash # removes all the nvidia drivers sudo apt-get --purge remove "*nvidia*" "libxnvctrl*" # reinstall the compatible driver and restart sudo ubuntu-drivers install ``` ### b) Purge CUDA from your computer > DO IT AT YOUR OWN RISK ```bash # removes all the nvidia drivers sudo apt-get --purge remove "*nvidia*" "libxnvctrl*" # remove all cuda versions sudo apt-get --purge remove "*cuda*" "*cublas*" "*cufft*" "*cufile*" "*curand*" "*cusolver*" "*cusparse*" "*gds-tools*" "*npp*" "*nvjpeg*" "nsight*" "*nvvm*" # remove all cuda folders sudo rm -rf /usr/loca/cuda* ``` ## Resources and helpful links - https://ubuntu.com/server/docs/nvidia-drivers-installation - https://developer.nvidia.com/cuda-toolkit-archive - https://developer.nvidia.com/cudnn-downloads -
Jul 14, 2023 . 1 changed file with 2 additions and 1 deletion.There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode charactersOriginal file line number Diff line number Diff line change @@ -15,7 +15,8 @@ This gist contains all the steps required to: ``` - Check GPU and available drives ```bash ubuntu-drivers devices # install it using: sudo apt install ubuntu-drivers-common ``` - Install the compatible driver -
Jun 7, 2023 . 1 changed file with 66 additions and 50 deletions.There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode charactersOriginal file line number Diff line number Diff line change @@ -1,11 +1,13 @@ # Steps to manage multiple CUDA environments This gist contains all the steps required to: - Install multiple CUDA versions (e.g., `CUDA 11.3` and `CUDA 11.8`). - Manage multiple CUDA environments on Ubuntu using the utility called [environment modules](https://modules.readthedocs.io/en/latest/). - Use this approach to avoid CUDA environment conflicts. > Environment Modules is a package that provides for the dynamic modification of a user's environment via modulefiles. You can find more on it at https://modules.readthedocs.io/en/latest/ ## 1. Install the Compatible NVIDIA Drivers (if required) - Add PPA GPU Drivers Repository to the System ```bash @@ -22,19 +24,20 @@ This gist contains all the steps required to sudo apt install nvidia-driver-530 ``` - Check the installed NVIDIA driver ```bash nvidia-detector ``` > **Note**: > - You can also auto-install the compatible driver using `sudo ubuntu-drivers autoinstall`. > - Additionally, you can also install NVIDIA drivers using the **Software & Updates** Ubuntu app. Just go to the **Additional Drivers** tab, choose a driver, and click **Apply Changes**. ## 2. Install `CUDA 11.3` and `CUDA 11.8` - Go to the [https://developer.nvidia.com/cuda-toolkit-archive](https://developer.nvidia.com/cuda-toolkit-archive) and select `CUDA Toolkit 11.3` from the available options. - Choose your OS, architecture, distribution, version, and installer type. For example, in my case: Option | value | :---:|:---:| @@ -44,7 +47,7 @@ This gist contains all the steps required to | Version | 20.04 | | Installer type | deb(local) | - Follow the provided installation instructions by copying and pasting the commands into your terminal. This will install `CUDA 11.3`. Use the following commands: ```bash wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin sudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600 @@ -54,7 +57,7 @@ This gist contains all the steps required to sudo apt-get update sudo apt-get -y install cuda ``` - Similarly, install `CUDA 11.8` using the following commands: ```bash wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin sudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600 @@ -64,17 +67,18 @@ This gist contains all the steps required to sudo apt-get update sudo apt-get -y install cuda ``` - Make sure to copy and execute the commands above in your terminal to install `CUDA 11.3` and `CUDA 11.8` on your system. ## 3. Install `cuDNN` library - Go to https://developer.nvidia.com/cudnn and download the `cuDNN` library for `CUDA 11.x`. Note that you might need to create a developer's account first. - Untar the downloaded file using the following command: ```bash tar -xvf cudnn-linux-x86_64-8.9.2.26_cuda11-archive.tar.xz ``` - Copy the `cuDNN` files to the `CUDA` toolkit files: ```bash # for CUDA 11.3 sudo cp include/cudnn*.h /usr/local/cuda-11.3/include @@ -85,35 +89,47 @@ This gist contains all the steps required to sudo cp lib64/libcudnn* /usr/local/cuda-11.8/lib64 ``` - Make the files executable: ```bash sudo chmod a+r /usr/local/cuda-11.3/include/cudnn*.h /usr/local/cuda-11.3/lib64/libcudnn* sudo chmod a+r /usr/local/cuda-11.8/include/cudnn*.h /usr/local/cuda-11.8/lib64/libcudnn* ``` > **Note**: > Strictly speaking, you are done with the CUDA setup. You can use it by adding the CUDA bin and library path to the PATH and LD_LIBRARY_PATH environment variables. For example, you can set up CUDA 11.8 by adding the following lines in the `~/.bashrc`: > ```bash > PATH=/usr/local/cuda-11.3/bin:$PATH > LD_LIBRARY_PATH=/usr/local/cuda-11.8/extras/CUPTI/lib64:$LD_LIBRARY_PATH > LD_LIBRARY_PATH=/usr/local/cuda-11.8/lib64:$LD_LIBRARY_PATH > ``` > Similarly, you can set up CUDA 11.3. However, manually changing the paths every time can be cumbersome! ## 4. Manage multipe CUDA versions using `environment modules` ### a) Install the environment modules utility: - Run the following commands: ```bash sudo apt-get update sudo apt-get install environment-modules ``` - Check the installation: ```bash # Check the installation by running module list ``` > You should see a list of default installed modules like git and maybe their versions displayed when you run the command `module list`. This confirms that the environment modules utility has been successfully installed on your system. ### b) Create modulefiles for CUDA distributions > **Note**: You might need root permissions to create directories and files. Use sudo in that case. - Create a directory `/usr/share/modules/modulefiles/cuda` to hold modulefiles for cuda distributions ```bash sudo mkdir -p /usr/share/modules/modulefiles/cuda ``` - Create a modulefile `/usr/share/modules/modulefiles/cuda/11.3` for `CUDA 11.3` and add the following lines: ```bash #%Module1.0 ## @@ -142,7 +158,7 @@ This gist contains all the steps required to conflict cuda ``` - Similarly, create a modulefile `/usr/share/modules/modulefiles/cuda/11.3` for `CUDA 11.3` and add the following lines: ```bash #%Module1.0 ## @@ -178,26 +194,26 @@ This gist contains all the steps required to set ModulesVersion 11.8 ``` > **Note**: make sure to reload your terminal. ## 5. Changing and Viewing the CUDA Module - To change and view the loaded CUDA module, you can use the following commands: ```bash # Check the currently loaded module module list # Check the available modules module avail # Load a specific cuda version module load cuda/11.3 # Unload the currently loaded CUDA module module unload cuda # Load CUDA 11.8 module load cuda/11.8 # verify the paths of the loaded CUDA nvcc --version # should give the loaded CUDA version echo $CUDA_HOME echo $PATH echo $LD_LIBRARY_PATH ``` > **Note**: You can add additional `CUDA` versions or other packages by creating corresponding modulefiles and following the steps outlined in this gist. -
Jun 2, 2023 . 1 changed file with 155 additions and 51 deletions.There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode charactersOriginal file line number Diff line number Diff line change @@ -1,78 +1,182 @@ # Steps to manage multiple CUDA environments This gist contains all the steps required to - Install multiple cuda versions (for eg. I install `CUDA 11.3` and `CUDA 11.8`) - Managing the multiple cuda environments on ubuntu using the utility called `environment modules` - Using such an approach avoids the cuda environment conflicts ## 1. Install the compatible nvidia drivers (if required) - Add PPA GPU Drivers Repository to the System ```bash sudo add-apt-repository ppa:graphics-drivers/ppa ``` - Check GPU and available drives ```bash ubuntu-devices drivers ``` - Install the compatible driver ```bash # in my case it is nvidia-driver-530 sudo apt install nvidia-driver-530 ``` - Check the installed nvidia driver ```bash nvidia-detector ``` > Note: > - You can also autoinstall the compatible using `sudo ubuntu-drivers autoinstall`. > - Additionally, you can also install nvidia-drivers using **Software & Updates** ubuntu app. Just go to **additional drivers** tab, choose a driver and click **apply changes**. ## 2. Install `CUDA 11.3` and `CUDA 11.8` - Go to the [https://developer.nvidia.com/cuda-toolkit-archive](https://developer.nvidia.com/cuda-toolkit-archive) and select `CUDA Toolkit 11.3.0` from the available options. - Now, select your OS, architecture, distribution, version and installer type. For example, in my case it is: Option | value | :---:|:---:| | OS | Linux | | Architecture | x86_64 | | Distribution | Linux | | Version | 20.04 | | Installer type | deb(local) | - You will get installation instructions, copy and paste in your terminal. This will install `CUDA 11.3` ```bash wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin sudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600 wget https://developer.download.nvidia.com/compute/cuda/11.3.0/local_installers/cuda-repo-ubuntu2004-11-3-local_11.3.0-465.19.01-1_amd64.deb sudo dpkg -i cuda-repo-ubuntu2004-11-3-local_11.3.0-465.19.01-1_amd64.deb sudo apt-key add /var/cuda-repo-ubuntu2004-11-3-local/7fa2af80.pub sudo apt-get update sudo apt-get -y install cuda ``` - Similarly, install `CUDA 11.8` ```bash wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin sudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600 wget https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda-repo-ubuntu2004-11-8-local_11.8.0-520.61.05-1_amd64.deb sudo dpkg -i cuda-repo-ubuntu2004-11-8-local_11.8.0-520.61.05-1_amd64.deb sudo cp /var/cuda-repo-ubuntu2004-11-8-local/cuda-*-keyring.gpg /usr/share/keyrings/ sudo apt-get update sudo apt-get -y install cuda ``` ## 3. Install `cuDNN` library - Go to https://developer.nvidia.com/cudnn and download the `cuDNN` library for CUDA 11.x > Note: you might need to make a developer's account first. - Untar the downloaded file ```bash tar -xvf cudnn-linux-x86_64-8.9.2.26_cuda11-archive.tar.xz ``` - Copy the cuDNN files to CUDA toolkit files ```bash # for CUDA 11.3 sudo cp include/cudnn*.h /usr/local/cuda-11.3/include sudo cp lib64/libcudnn* /usr/local/cuda-11.3/lib64 # for CUDA 11.8 sudo cp include/cudnn*.h /usr/local/cuda-11.8/include sudo cp lib64/libcudnn* /usr/local/cuda-11.8/lib64 ``` - Make the files executable ```bash sudo chmod a+r /usr/local/cuda-11.3/include/cudnn*.h /usr/local/cuda-11.3/lib64/libcudnn* sudo chmod a+r /usr/local/cuda-11.8/include/cudnn*.h /usr/local/cuda-11.8/lib64/libcudnn* ``` > Note: > Strictly speaking, you have the CUDA setup now. You can use it by adding the CUDA bin and library path to `PATH` and `LD_LIBRARY_PATH` environment variables ## 4. Manage multipe cuda version using `environment modules` ### a) Install the environment modules utility ```bash sudo apt-get update sudo apt-get install environment-modules ``` ```bash # Check the installation by running module list ``` ### b) Create modulefiles for cuda distributions > **Note**: you might need `root` permissions to make directory and create files, use `sudo` in that case - Create a directory `/usr/share/modules/modulefiles/cuda` to hold modulefiles for cuda distributions - Create a modulefile `/usr/share/modules/modulefiles/cuda/11.3` for `CUDA 11.3` and following lines to the file: ```bash #%Module1.0 ## ## cuda 11.3 modulefile ## proc ModulesHelp { } { global version puts stderr "\tSets up environment for CUDA $version\n" } module-whatis "sets up environment for CUDA 11.8" if { [ is-loaded cuda/11.8 ] } { module unload cuda/11.8 } set version 11.3 set root /usr/local/cuda-11.3 setenv CUDA_HOME $root prepend-path PATH $root/bin prepend-path LD_LIBRARY_PATH $root/extras/CUPTI/lib64 prepend-path LD_LIBRARY_PATH $root/lib64 conflict cuda ``` - Create a modulefile `/usr/share/modules/modulefiles/cuda/11.8` for `CUDA 11.8` and following lines to the file: ```bash #%Module1.0 ## ## cuda 11.8 modulefile ## proc ModulesHelp { } { global version puts stderr "\tSets up environment for CUDA $version\n" } module-whatis "sets up environment for CUDA 11.8" if { [ is-loaded cuda/11.3 ] } { module unload cuda/11.3 } set version 11.8 set root /usr/local/cuda-11.8 setenv CUDA_HOME $root prepend-path PATH $root/bin prepend-path LD_LIBRARY_PATH $root/extras/CUPTI/lib64 prepend-path LD_LIBRARY_PATH $root/lib64 conflict cuda ``` ### c) Make `CUDA 11.8` the default cuda version - Create a file `/usr/share/modules/modulefiles/cuda.version` to make `CUDA 11.8` the default cuda module: ```bash #%Module set ModulesVersion 11.8 ``` ## Quick examples to switch cuda versions ```bash @@ -96,4 +200,4 @@ which nvcc echo $CUDA echo $PATH echo $LD_LIBRARY_PATH ``` -
Jun 1, 2023 .There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode charactersOriginal file line number Diff line number Diff line change @@ -0,0 +1,99 @@ # Steps to manage multiple CUDA environments This gist contains all the steps required to manage multiple cuda environments on ubuntu. For the steps walkthrough, For the walkthrough, I will install `CUDA 11.8` and `CUDA 11.3` on my system and manage them using `environment modules`. ## 1. Install the compatible nvidia drivers (if required) - Add PPA GPU Drivers Repository to the System ```bash sudo add-apt-repository ppa:graphics-drivers/ppa ``` - Check GPU and available drives ```bash ubuntu-devices drivers ``` - Install the compatible driver ```bash sudo apt install nvidia-driver-530 ``` - Check the installed nvidia driver ```bash nvidia-detector ``` - Note: > - you can also autoinstall the compatible using `sudo ubuntu-drivers autoinstall` > - Additionally, you can also install nvidia-drivers using **Software & Updates** ubuntu app. Just go to **additional drivers** tab, choose a driver and click **apply changes** ## 2. Install `CUDA 11.3` and `CUDA 11.8` ## 3. Install `cuDNN` library ## 4. Manage multipe cuda version using `environment modules` ### a) Install the environment modules utility ```bash sudo apt-get update sudo apt-get install environment-modules ``` ### b) Create modulefiles for cuda distributions - Create a directory `/etc/modulefiles/cuda` to hold modulefiles for cuda distributions ```bash mkdir /etc/modulefiles/cuda ``` - Create a modulefile /etc/modulefiles/cuda/11.8 for `CUDA 11.3` and following lines to the file: ```bash vim /etc/modulefiles/cuda/11.3 ``` ```bash lorem epsum ``` - Create a modulefile /etc/modulefiles/cuda/11.3 for `CUDA 11.8` and following lines to the file: ```bash vim /etc/modulefiles/cuda/11.8 ``` ```bash lorem epsum ``` ### c) Make `CUDA 11.8` the default cuda version - Create a file `/etc/modulefiles/cuda/.version` to make python/cuda/11.8 the default cuda module: ```bash lorem epsum ``` ## Quick examples to switch cuda versions ```bash ## check the path to nvcc (cuda distribution) which nvcc echo $CUDA_HOME echo $PATH echo $LD_LIBRARY_PATH # check the available modules module avail # list the loaded modules module list # load CUDA 11.3 module load cuda/11.3 # check loaded cuda distribution (should show CUDA 11.3 paths) which nvcc echo $CUDA echo $PATH echo $LD_LIBRARY_PATH ```