Skip to content

Summary

I can submit jobs that use a GPU.

#SBATCH -p node
#SBATCH -N 1
#SBATCH -C gpu
#SBATCH --gpus-per-node=2

I can transfer files to/from Bianca using rsync.

ssh sven@transit.uppmax.uu.se
mount_wharf sens2025560
rsync my_local_file.txt sven@transit.uppmax.uu.se:sens2025560
rsync --recursive my_folder sven@transit.uppmax.uu.se:sens2025560

I can see the CPU and memory usage of jobs

jobstats --plot 1234567

I understand how to set up jobs efficiently

  • Enough cores for memory
  • CPU limited? Consider adding more until CPU usage is around 80% on average
  • Consider adding 1 core for safety

I understand how to install software myself

  • Julia:
    • install locally with using <package-name>
    • copy ~/.julia/packages to Bianca
  • Python pip:
    • install locally with pip install
    • copy ~/.local/lib/python<version>/site-packages/ to Bianca

  • Python conda:

    module load conda
export CONDA_ENVS_PATH=/proj/sens2025560/$USER
conda create -c conda-forge --prefix $CONDA_ENVS_PATH/bianca-course python=3.7 numpy=1.15
source activate bianca-course
pip list
python -V
conda deactivate

  • R:

    • install locally with install.packages
    • copy ~/R/x86_64-pc-linux-gnu-library/4.2 to Bianca

I understand what containers are.

  • Apptainer
  • Docker
  • Singularity

I can can run the voted-for IDE on Bianca

  • Jupyter:
module load python/3.11.4
jupyter-notebook --ip 0.0.0.0 --no-browser
  • RStudio:
nteractive -A sens2025560 -n 2 -t 8:00:00
module load R/4.3.1 R_packages/4.3.1 RStudio/2023.06.2-561
rstudio
  • VSCodium:
module load VSCodium/latest
code