Selection Guide

Selection Guide#

*A Practical Guide to Selecting a Compute Environment

Choosing the right compute environment depends on two primary questions:

  1. How do you want to interact with your computation?

  2. How large and resource-intensive is your problem?

Step 1 — Interaction Style

Ask yourself:

  • Do I need to interact with the computation while it runs?

  • Am I developing or debugging code?

  • Do I want notebooks, plotting, or scripting in real time?

If yes → start with an interactive environment.

If no → a batch workflow is likely appropriate.

Step 2 — Scale and Resources

Next, consider resource requirements:

  • Fits within modest CPU and memory limits → Interactive environments are often sufficient

  • Requires long runtimes, many cores, or MPI → Batch HPC execution is required

A Simple Decision Guide#

  • I’m exploring, learning, or prototyping → JupyterHub

  • I want to run a small serial script with minimal setup, (e.g. OpenSees) → OpenSees-Express VM

  • I need large-scale, long-running, or parallel execution → HPC batch jobs on TACC systems

Choosing the Optimal Compute Environment#

Environment

Access Method

Resources & Scaling

Best For

JupyterHub

Browser (notebooks/terminal)

Up to 8 cores / 20 GB RAM per container

Prototyping, OpenSeesPy, quick test loops, automation

OpenSees-Express VM

Web Portal (form-based)

Shared lightweight VM

Tcl scripts, small serial jobs, easiest point of entry

HPC Environment

SSH + SLURM (Tapis)

Full system access, massive scale

Parallel OpenSees jobs, large simulations, production

A Typical Compute Path#

Many users naturally follow this progression:

  1. Develop and test interactively

  2. Validate on small problems

  3. Scale up using batch jobs

  4. Automate or repeat studies at scale

This transition is expected — and supported — within the DesignSafe ecosystem.

Why This Matters#

Understanding where your computation runs — and how it is executed — helps you:

  • Avoid unnecessary queues

  • Use resources efficiently

  • Scale smoothly from prototypes to production

  • Reduce friction as projects grow

Contents