Preview

ComputeDeck

A dashboard for jobs,
queues and GPUs.

Uses your SSH aliases and cloud accounts. No cluster administrator access needed.

Watch demo · 20s Features

macOS & Linux · Read-only collection · English & 中文

OVERVIEW EXAMPLE DATA

Overview of fictional clusters, cloud instances and GPU usage

Clusters, cloud instances and SSH machines.

20-second demo

Overview → queued job → GPU options

Example data · English AI voiceover and captions

Download MP4

Transcript

Meet ComputeDeck. See your clusters and cloud GPUs in one place. Open a queued job to understand the wait. Then choose your GPU needs and compare capacity, permissions, and time limits. Read-only by default, with your existing SSH setup. This walkthrough uses example data.

Jobs and capacity

Inspect a queued job

Open a job to see its GPU request, account, QoS and the scheduler’s pending reason.

Details of a fictional queued job requesting four A100 GPUs — Queued job details · example data

Compare GPU options

Enter a GPU model, count and run time. Compare visible capacity and verified QoS limits.

Estimates use the last collection. Reservations and priorities can still delay a job.

Options for four H100 GPUs over 24 hours, with capacity and run-time limits — 4 × H100 for 24 hours · example data

Supported sources

Collect on your laptop, or run on a server and connect through SSH.

Clusters: Slurm · Kubernetes · PBS · LSF
Cloud GPUs: RunPod · Lambda · AWS EC2 · Vast.ai
Other machines: GPU hosts over SSH · Manual entries

Kubernetes, PBS, LSF and Vast.ai integrations are beta.