Différences

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

--- public:support:labex-efl-gpu:lipn-a40 [2023/07/11 14:04]
garciaflores créée
+++ public:support:labex-efl-gpu:lipn-a40 [2023/07/11 14:44] (Version actuelle)
abedelkader [Installation of Pytorch library on the lipn-a40 server]
@@ Ligne 2: / Ligne 2: @@
 ===== Installation of Pytorch library on the lipn-a40 server =====
+''** If you had an problems with th CUDA version : **''
+''** You should follow these steps:**''
+  - You should remove the your environment using this command : <blockquote>rm -r miniconda3</blockquote>
+  - Then downlaod the miniconda using this command : <blockquote>wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh</blockquote>
+  - To install the miniconda , you willl have to change the permissions, since you are not a admin ( and you are not :) ) using this command you will add the the executable permission to your file using this command : <blockquote>chmod +x Miniconda3-latest-Linux-x86_64.sh</blockquote>
+  - To launch your environement you have to use this command : <blockquote>source ~/.bashrc</blockquote> or simply  <blockquote>conda activate</blockquote>
+  - For accessing the LIPN A40 GPUs you will use this command : <blockquote>ssh lipn-gpus-a40</blockquote>
+  - For installing the right tirhc version with the CUDA version you will use this command : <blockquote>pip3 install torch torchvision torchaudio</blockquote>
+  - For testing if you get the right acess for the GPUs use this command : <blockquote>srun python3 gpu_test.py</blockquote>
+**IT'S VERY IMPORTANT TO FOLLOW THESE STEPS IN THE RIGHT ORDER**