HPC Dashboard Download

Resource insight for research computing

See what your HPC jobs actually used.

Your scheduler knows what you requested. HPC Dashboard shows what happened: where your cores, memory and GPUs went, while the job runs and after it finishes.

Download the app How it works

Free desktop app · macOS, Windows & Linux · Gadi / PBS today

01Live & finished jobs
02Per-core visibility
03Local desktop app
04No cluster-side service

THE QUESTION BEHIND THE BILL

37% CPU efficiency.
Now what?

A single average cannot tell you what to change. Were some cores fully busy while others sat idle? Or was every core waiting on I/O? Those lead to very different next runs.

Explore the advice guide
A / IDLE CORES

A smaller request may help.

Some cores do the work. The rest barely move.

B / EVERY CORE WAITS

Find the bottleneck first.

All cores are active at a similar, lower level.

Same job-level average. Different decisions.

FROM TELEMETRY TO A DECISION

See the whole run.
Understand the details.

Read your job at a glance, then follow the evidence down to individual cores and nodes.

01 / LIVE VIEW

Watch jobs as they run

Follow CPU, memory, disk, network and GPU metrics without waiting for the final report.

02 / CORE DETAIL

Find the idle cores

See how work was distributed across cores and nodes, beyond a single average.

03 / NEXT RUN

Get practical advice

Turn usage into a better resource request, with NCI charging rules in mind.

04 / COMPARE

Compare two runs

Check whether a code or configuration change made a measurable difference.

A SHORT PATH TO YOUR FIRST CHART

Set it up once.
Learn from every run.

The app reads telemetry files over your SSH connection or a local mount. It does not submit, cancel or modify jobs.

01

Collect telemetry in your job

Load logger-rs and mark the workload you want measured in your PBS script.

02

Connect the desktop app

Download HPC Dashboard and add a profile for the directory where your logs land.

03

Open a job and explore

Inspect live charts, read the finished summary, compare runs and plan your next request.

job.pbsFull setup guide 
# Inside your PBS job script
module load logger-rs
source logger-rs.sh

telemetry_start
mpirun -np "$PBS_NCPUS" ./my_model
telemetry_stop

Module paths and storage directives vary by site. See the guide before your first run.

READY TO SEE THE FULL PICTURE?

Make the next run
a better one.

Download the latest release, then follow the quick setup guide to start collecting useful evidence from your jobs.

Download for your platform Read the documentation

Available for macOS, Windows and Linux. Check the installation guide for current signing notes.