Différences

Ci-dessous, les différences entre deux révisions de la page.

Lien vers cette vue comparative

Les deux révisions précédentes Révision précédente
Prochaine révision
Révision précédente
deeplearning:dgx [2019/01/25 10:05] – [Anaconda] admindeeplearning:dgx [2020/02/25 13:57] (Version actuelle) – [Anaconda] admin
Ligne 3: Ligne 3:
  
  
-The DGX station is opened only for allowed users. For security reasons it is isolated from the rest of cluster.  +<note important>The DGX station is opened only for allowed users (**I-SITE ADVANCES**). For security reasons it is isolated from the rest of cluster.  
 +</note>
 <note> <note>
    * This node is shared between users and is not managed by a resources manager (GridEngine).  Thank you to not monopolize resources    * This node is shared between users and is not managed by a resources manager (GridEngine).  Thank you to not monopolize resources
Ligne 61: Ligne 61:
 </note> </note>
 ===== Using default installed software ===== ===== Using default installed software =====
-==== Vanilla python ====+==== Vanilla Python ====
  
 Several programs are installed by default ''python2.*'' and ''python3.*'' Several programs are installed by default ''python2.*'' and ''python3.*''
Ligne 88: Ligne 88:
 <code bash> <code bash>
 [user@mesodgx~]$ python my_python_program.py      [user@mesodgx~]$ python my_python_program.py     
 +</code>
 +You can install new package **without** root access. For example:
 +<code bash>
 +$ pip install --user package
 </code> </code>
  
Ligne 93: Ligne 97:
 For better performances Anaconda python V3 is installed in ''/opt/''. Several optimized scientific packages are available.  For better performances Anaconda python V3 is installed in ''/opt/''. Several optimized scientific packages are available. 
  
-To use anaconda, please add this line to your ''.bashrc'' file+ 
 + 
 + 
 +To use anaconda, please add these lines to your ''.bashrc'' file
 <code bash>  <code bash> 
 export PATH="/opt/anaconda3/bin:$PATH" export PATH="/opt/anaconda3/bin:$PATH"
 +. /opt/anaconda3/etc/profile.d/conda.sh
 </code> </code>
 +
 +
 Check for available packages: Check for available packages:
 <code bash> <code bash>
Ligne 121: Ligne 131:
 You may need to install additional package without **having** root access. We instead recommend building a ''virual enviroment'' locally and install packages there. You may need to install additional package without **having** root access. We instead recommend building a ''virual enviroment'' locally and install packages there.
   - Create a conda environment <code bash>$ conda create -y -n conda-env</code> replace ''conda-env'' by a valid name.    - Create a conda environment <code bash>$ conda create -y -n conda-env</code> replace ''conda-env'' by a valid name. 
-  - Activate the environment you just created <code bash>source activate conda-env </code> +  - Activate the environment you just created <code bash>$ source activate conda-env </code> 
-  - Download and install the conda package in the environment <code bash> $conda install -c package</code> replace package by the package you wan to install, for example ''panda''+  - Download and install the conda package in the environment <code bash> $ conda install -c package</code> replace package by the package you wan to install, for example ''panda''
   - Run your analysis from **within** the environment   - Run your analysis from **within** the environment
-  - Deactivating environment, Oonce you are done with your analysis- you can deactivate the conda environment by running the command <code bash>$ source deactivate</code>+  - Deactivating environment, once you are done with your analysis- you can deactivate the conda environment by running the command <code bash>$ source deactivate</code> 
 + 
 +**Example: installing Tensorflow GPU** 
 + 
 +<code bash> 
 +$ conda create -y -n tensorflowGPU 
 +</code>  
 +<code bash> 
 +$ conda activate tensorflowGPU 
 +</code> 
 +<code bash> 
 +$ (tensorflowGPU) user@mesodgx:~$ conda install -c anaconda tensorflow-gpu  
 +</code> 
 +<code python> 
 +$ (tensorflowGPU) user@mesodgx: python 
 +Python 3.6.8 |Anaconda, Inc.| (default, Dec 30 2018, 01:22:34)  
 +[GCC 7.3.0] on linux 
 +Type "help", "copyright", "credits" or "license" for more information. 
 +>>> import tensorflow as tf 
 +>>> sess = tf.Session(config=tf.ConfigProto(log_device_placement=True)) 
 + 
 +</code> 
 + 
 +<note tip>Use <code bash>conda info --envs</code> to view all your environment</note>
 ===== Using Docker ===== ===== Using Docker =====
 Before pulling the images, check the available images already downloaded. Before pulling the images, check the available images already downloaded.
Ligne 320: Ligne 353:
 {{ :deeplearning:visu_volta.png?600 |}} {{ :deeplearning:visu_volta.png?600 |}}
 */ */
 +======= Tutorial and Labs =======
 +  * [[deeplearning:docker| Using docker and tensorflow]]
 +
 ======= Links ======= ======= Links =======
   * https://dgx-wiki.readthedocs.io/en/latest/docs/docker/index.html   * https://dgx-wiki.readthedocs.io/en/latest/docs/docker/index.html
   * https://devblogs.nvidia.com/docker-compatibility-singularity-hpc/   * https://devblogs.nvidia.com/docker-compatibility-singularity-hpc/
   * https://docs.nvidia.com/deeplearning/dgx/index.html   * https://docs.nvidia.com/deeplearning/dgx/index.html
 +  * https://conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html
 +  * https://docs.nvidia.com/dgx/dgx-station-user-guide/index.html
 +  * https://docs.nvidia.com/dgx/ngc-registry-for-dgx-user-guide/index.html
  • deeplearning/dgx.1548410748.txt.gz
  • Dernière modification : 2019/01/25 10:05
  • de admin