Quick Reference

Last updated on 2026-08-27 | Edit this page

Sagehen HPC Quick Reference Card

Connection & Access

Command Purpose
ssh <myusername>@sagehen.hpc.pomona.edu Connect via SSH
exit or logout Disconnect from Sagehen
https://ondemand.hpc.pomona.edu/ Web portal (no SSH needed)

Cluster Information

Command Purpose
sinfo Show all partitions and node status
sinfo -l Detailed partition information
sinfo -N Show all nodes
sinfo -p amd Show specific partition (e.g., amd)
lscpu Show CPU information
free -h Show available memory

File & Directory Navigation

Command Purpose
pwd Print working directory
ls List files
ls -lah Long format with hidden files
cd ~ Go to home directory
cd directory Change to directory
cd .. Go up one level
mkdir dirname Create directory
mkdir -p path/to/dir Create nested directories
rm file Delete file
rm -r directory Delete directory and contents

Storage Locations

Path Purpose Backup Size
/rhome/<myusername> Home directory Yes ~100 GB
/bigdata/lab/<labname> Lab shared storage Yes 1 TB
/scratch Temporary job storage No No quota*
/tmpfs RAM disk (during jobs) No ~1 GB

*Deleted when your job completes – copy results out before the job ends

Checking Storage Usage

Command Purpose
quota_check.sh Usage vs. quota (use this, not plain du)
du -sh --apparent-size ~ Approximate home directory size
du -sh --apparent-size /bigdata/lab/<labname> Approximate lab storage size

Note: on BeeGFS, plain du does not report quota usage correctly – use quota_check.sh.

Module System (Lmod)

Command Purpose
module avail List all available modules
module avail miniconda3 Search for modules (e.g., miniconda3)
module load miniconda3 Load default version
module load miniconda3/py313_26.3.2-2 Load specific version
module unload miniconda3 Unload module
module purge Unload all modules
module list Show currently loaded modules
module show miniconda3 Show module details

Job Submission & Management

Command Purpose
sbatch script.sh Submit batch job
squeue -u $USER Check your job status
squeue -j JOBID Check specific job
scancel JOBID Cancel a job
sacct -u $USER Show job history
sacct -j JOBID Details of specific job

Common SBATCH Directives

BASH 

#!/bin/bash
#SBATCH --job-name=myname        # Job name
#SBATCH --partition=short         # Partition (short, amd, gpu)
#SBATCH --time=01:00:00          # Max runtime HH:MM:SS
#SBATCH --ntasks=1               # Number of tasks
#SBATCH --cpus-per-task=4        # Cores per task (1-128)
#SBATCH --mem=8G                 # Memory per job
#SBATCH --output=job_%j.log      # Output file (%j = job ID)
#SBATCH --gres=gpu:1             # Request 1 GPU

File Operations

Command Purpose
cat file.txt View file contents
head -n 10 file.txt Show first 10 lines
tail -n 10 file.txt Show last 10 lines
grep "pattern" file.txt Search for text in file
wc -l file.txt Count lines
chmod +x script.sh Make file executable
cp source dest Copy file
mv source dest Move/rename file

Text Editors

Command Purpose
nano file.txt Simple editor
vim file.txt Advanced editor

Job Script Template

BASH 

#!/bin/bash
#SBATCH --job-name=myanalysis
#SBATCH --partition=amd
#SBATCH --time=04:00:00
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=8
#SBATCH --mem=16G
#SBATCH --output=%j_output.log

# Load modules needed
module purge
module load miniconda3/py313_26.3.2-2

# Run your analysis
echo "Starting analysis..."
python my_script.py
echo "Analysis complete!"

Partition Summary

Partition Max Time Cores GPUs Best For
short Shorter – check sinfo -p short 128 0 Testing & debugging
amd 30 days 128 0 General compute
gpu 30 days 128 Yes (per-account limits) GPU-accelerated work

Environment Variables

Variable Purpose
$USER Your username
$HOME Your home directory
$SLURM_JOB_ID Current job ID
$SLURM_JOB_NAME Current job name
$SLURM_CPUS_PER_TASK Cores allocated
$SLURM_JOB_PARTITION Current partition
$SCRATCH Scratch directory (during jobs)

Useful Aliases (Add to ~/.bashrc)

BASH 

alias ll='ls -lah'
alias sq='squeue -u $USER'
alias sa='sacct -u $USER'
alias cd-home='cd ~'
alias du-home='du -sh ~'

Common Error Messages

Error Solution
sbatch: command not found You may be on a compute node or wrong host; run from the head node, or contact
Permission denied (publickey) Wrong username or password
Job timeout Increase --time in sbatch script
Out of memory Increase --mem in sbatch script
Module not found Check spelling; use module avail to find correct name

Contact & Support

Issue Contact
HPC account problems
Technical issues
Sagehen access
Pomona IT general (909) 621-8061 /
This workshop Ask your instructor

Sagehen HPC Hardware Summary

Compute Nodes (amd partition): - 12 nodes (a001-a012) - 128 cores, 500 GB RAM each - 30-day max job time

GPU Nodes (gpu partition): - 10 GPUs total across the GPU nodes - 128 cores, 500 GB RAM each - 4x A100 (80 GB), 4x L40S (48 GB), 2x RTX PRO 6000 (96 GB)

Head Node (sagehen.hpc.pomona.edu): - 2 cores, 8 GB shared RAM - Don’t run jobs here! - Use for: login, submission, compilation

  1. Test job on short partition (fast feedback)
  2. Debug with squeue and output files
  3. Fix issues and increase --time for longer runs
  4. Submit to amd partition for production
  5. Monitor with squeue and sacct
  6. Copy results from scratch to /rhome or /bigdata
  7. Clean up temporary files before quota limit

More Information