SSH:

login: username@ada.qtinslab.com

SLURM (Recommended)

Job submission: sbatch run.shell
Job queue: squeue

run.shell may read

#!/bin/bash

#SBATCH --job-name=test_two_gpus

#SBATCH --output=test_two_gpus_%j.out

#SBATCH --error=test_two_gpus_%j.err

#SBATCH --partition=main

#SBATCH --gres=gpu:2 (max 2)

#SBATCH --cpus-per-task=8 (max 112)

#SBATCH --mem=32G (max 128G)

#SBATCH --time=01:00:00 (max INFINITY)

#----* Feel free to add other SBATCH *----

echo "Running on host: $(hostname)"

echo "GPUs allocated:"

nvidia-smi

# -- * OPTIONAL * --

# Activate your virtual environment

# source /opt/anaconda3/lib/python3.13/site-packages/conda/shell/etc/profile.d/conda.sh

# conda activate torch_env_3.12.4

python small_net.py

Documentations: Slurm Workload Manager - Quick Start User Guide 

TASK-SPOOLER

Private Socket
Private job submission/queue: ts

Examples:
ts python x.py (cpu only)
ts -G 2 python x.py (2 GPUs)
ts: view job’s status and many things.

Useful commands: man ts