T
Telemetry
Live GPU telemetry
Utilization, memory, temperature, and power for all four L40S cards, refreshed continuously. You see the cluster as it is, not as it was when you last ran nvidia-smi.
Version
0.1.0
Software
Tools we build for our own lab, shared.
AuraPulse-S is the desktop client we wrote for the AURA Lab's GPU server — built because our students needed a sane way to see the cluster, submit jobs, and read their logs. It is free for W&M CS students.
01
The app
One window onto the GPU server.
AuraPulse-S connects to the AURA Lab's GPU server — four NVIDIA L40S GPUs — over SSH. Telemetry, queue state, and your terminal live in a single native window — no browser tabs, no copy-pasted hostnames, no forgotten squeue flags.
J
Jobs
SLURM submission & monitoring
Submit jobs from the app, watch them move through the queue, and tail their logs live. Pick 1–4 GPUs, set your time limit, and AuraPulse-S writes the sbatch script for you.
S
Scope
Scoped to you
The app authenticates as you and only ever shows your jobs, your files, your allocation. Credentials live in the macOS Keychain; nothing is proxied through a third party.
>_
Terminal
Built-in SSH terminal
A full terminal session to the GPU server inside the app, for everything the GUI doesn't cover. Same connection, same credentials, one less window to manage.
02
Download
Get AuraPulse-S.
Free for W&M CS students.
Requires macOS 14 Sonoma or later on Apple Silicon. On first launch, right-click the app and choose Open — this build is self-signed, so Gatekeeper will ask once. A notarized release is on the way.
Requirements
- macOS 14 Sonoma or later
- Apple Silicon (M1 and newer)
- A W&M CS account with access to the AURA GPU server
- Free for W&M CS students
03
First run
Three steps to your first job.
01
Connect
Open & connect
The connection wizard has the AURA GPU server prefilled. Sign in with your W&M id; the password goes into the macOS Keychain and stays there.
02
Personalize
Make it yours
Pick a theme, set the launch sequence, and arrange the panels the way you work. It remembers everything.
03
Submit
Submit your first job
Jobs → Submit job. Choose 1–4 GPUs, watch the queue from the dashboard, and tail the logs live while it runs.
Questions, bugs, or feature requests — email amastropaolo@wm.edu. If the server is down, the app will tell you before we do.