In your terminal:
ssh your-username@glogin.ibex.kaust.edu.sa |
module load singularity |
I am assuming that you have registered with NVIDIA NGC and created an API key which is required to access NGC image registry. If not, please follow this page to do so:
https://docs.nvidia.com/ngc/ngc-overview/index.html#registering-activating-ngc-account
To pull the latest image from NGC registry for Kaldi you can invoke the following command:
cd $HOME export SINGULARITY_TMPDIR=$HOME singularity pull docker://nvcr.io/nvidia/kaldi:20.11-py3 |
This may take a few 10s of minutes.
The image will be downloaded and be converted to a format that Singularity understands, called SIF file or Singularity Image File. Once you have a SIF file you are ready to run the container.
salloc --reservation=hackathon2020 --gres=gpu:v100:1 --time=01:00:00 module load singularity srun --pty singularity shell --nv kaldi_20.11-py3.sif |
Jobscript
#!/bin/bash #SBATCH --gres=gpu:v100:1 #SBATCH --ntasks=1 #SBATCH --time=00:30:00 module load singularity singularity run --nv kaldi:20.11-py3 your_application |