Filepaths and Useful tricks

This page should be updated as filepaths that the lab uses change over time and as people develop useful tricks that are worth sharing.

Common engaging filepaths

• Dataset folders: /orcd/data/satra/002/datasets/
• Project folders: /orcd/data/satra/002/projects/
• User folder: /orcd/data/satra/001/users/$USER/
• Creating scratch results: /orcd/scratch/bcs/001/$USER/

.bashrc Recommendations

Useful aliases

• Starting a non-gpu compute node: alias sr='srun -n 1 -t 08:00:00 --mem=20GB -p mit_normal --exclude node2704 --pty /bin/bash' [TODO: it is likely possible to set this command to accept values especially for number of CPUs, memory and time needs]
• Starting a GPU compute node: alias sr_gpu='srun -n 4 -N 1 -t 4:00:00 --mem=40GB -p ou_bcs_high,pi_satra --gres=gpu:1 --pty bash'
• Viewing active nodes/jobs: alias sq='squeue -u $USER' (it's possible that replacing $USER with your username might speed this command up, though it is likely minimal)
• Navigating quickly to some common directories:
  • alias cd_datasets='cd /orcd/data/satra/002/datasets/'
  • alias cd_projects='cd /orcd/data/satra/002/projects/'
• Alternatively to the above, you could replace the absolute file paths with environment variables (see next section)

Common environment variables

• For dataset folder/or a specific dataset: export SIG_DATASETS_DIR='/orcd/data/satra/002/datasets/' (change the name and add a dataset after if you find you commonly need to reference it, especially within code)
• Similar for projects: export SIG_PROJECTS_DIR='/orcd/data/satra/002/projects/'