⚙️ Best Practices for Using HPC Cluster

Follow these best practices to ensure efficient, secure, and optimal usage of HPC resources.

1. ❌ Do NOT run long jobs on login nodes.
   ✔ Use login nodes only for compilation and light tasks.
2. Refer to the Quick Start Guide for initial setup and guidance.
3. 📄 Sample Job Scripts: Access ready-to-use job submission scripts from the Job Submission Guide .
4. Use the same compiler across all modules/libraries to avoid compatibility issues.
5. Optimize performance using appropriate compiler flags (e.g., -O3) and verify output accuracy.
6. Ensure modules/libraries used during execution match those used during compilation.
7. Monitor disk usage before submitting multiple jobs to avoid storage issues.
8. ✔ Submit jobs from $SCRATCH
   ✔ Store final results in $HOME
9. ⚠️ $SCRATCH is NOT backed up. Always download important data to local systems.
10. Install software only from trusted sources to avoid malware/ransomware risks.
11. Avoid spaces in file and directory names (use underscores instead).
12. Report unusual behavior (slowdowns, missing/corrupt files) to support immediately.
13. Use GPU and large partitions (gpu, large) only when required.
14. Specify wall-time carefully in job scripts — it directly affects queue wait time.