slurm
- Commandsinfo
scontrol show node [node_tag]
ssh [node_tag]
srun
# Command
sbatch \
--comment [comment] \
--timeout [max_execution_time] \
--exclusive \
--nodes [node_number] \
--cpus-per-task [cpu_number] \
--wrap [command]
# Example
sbatch \
--comment "feicheng" \
--timeout 60000 \
--exclusive \
--nodes 32 \
--cpus-per-task 128
--wrap "./run_inference_demo.sh"
squeue
scancel [task_id]