PACE

Intro

PACE at Georgia Tech

The Partnership for an Advanced Computing Environment (PACE) provides leading-edge high performance computing (HPC) resources with technical support services, infrastructure, software, and mor

• See PACE Environment Setup for logging into Phoenix Cluster using SSH

Pace Onboarding

Access PACE clusters through a web browser

Using the web interface is usesful for requesting PACE resources

Run GUI Applications on the Cluster¶

Open OnDemand (also known as OnDemand) is an interface that allows you to access PACE clusters through a web browser

Run browser-based VS Code session on a compute node

• Visit My Interactive Sessions in the Phoenix Dashboard to start a browser-based VS Code session on a compute node

Configure Session

• Set charge account

paceship-dsgt_clef2026

• Set QoS

  • CPU (free): embers
  • GPU ($1K credit): inferno

• Set Node type (e.g. CPU-only, Intel or NVIDIA GPU RTX6000)

• Click Launch → Connect to open VS Code in the browser

• Click Open Folder → /storage/home/hcoda1/1/username/

Login to PACE using SSH

• See Log in to PACE using SSH

Informational Commands ¹

• Check id of compute node

hostname

atl1-1-02-018-19-1.pace.gatech.edu

• Connect to the compute node (id will change whenever you request a new resource)

ssh atl1-1-02-018-19-1

• Show PACE folders

ls -la

• See all pending and running jobs in the queue

squeue

• See resources you are using

squeue -u username

• Show storage breakdown
  • Home: 20 GB
    • For code/repos
  • Projct: 1TB
    • Data for downloading sharing with team
  • Scratch: 15TB
    • Personal use, install dependencies/virtual environments, gets deleted after 90 days of inactivity

pace-quota

Footnotes

1. https://gatech.service-now.com/technology?id=kb_article_view&sysparm_article=KB0041998#informational-commands ↩