Repolode
Submit
MiaAI-Lab

MiaAI-Lab/sparkDash

A self-hosted real-time web dashboard that monitors one or more NVIDIA DGX Spark and other GPU machines (GPU, memory, CPU, network, local LLM servers) in one browser window.

View on GitHubWebsite
Stars
838
Forks
201
Open issues
16
Last commit
yesterday

What is it?

sparkDash is a web dashboard for watching several NVIDIA DGX Spark (GB10) machines at once. It streams GPU, CPU, unified memory, storage, network and local LLM metrics over WebSocket, and units can be added, edited, reordered or removed from the UI without restarting the server. Linux machines with a discrete NVIDIA GPU can be added as dedicated GPU hosts over SSH, with system RAM and VRAM shown separately. Windows PCs with NVIDIA or AMD iGPUs and macOS machines are also supported as monitored units.

On the LLM side, sparkDash auto-detects common inference servers such as llama.cpp, vLLM, SGLang, LM Studio and Ollama, and reports decode and prefill tokens per second where the engine publishes them. It also includes benchmark pages for decode, prefill and quality, a token and energy history, health findings that are logged to an activity feed, and an opt-in Prometheus `/metrics` endpoint. Optional probes cover ComfyUI jobs, Hermes Agent updates and Tailscale connectivity.

The project is Docker-first, runs as a single container, and stores SSH passwords encrypted with AES-256-GCM. It ships as an installable PWA, a `/kiosk` wall-display mode, and is written in JavaScript with React and Express.

Key features

  • 01Real-time GPU, CPU, memory, storage and network metrics streamed to the browser over WebSocket.
  • 02Add, edit, reorder or remove units from the UI without restarting.
  • 03Auto-detects common LLM servers and reports decode and prefill tok/s where published.
  • 04Decode, prefill and quality benchmark pages with history.
  • 05Health findings such as GPU overheating and Xid or OOM errors, logged to an activity feed.
  • 06Opt-in Prometheus `/metrics` export for external monitoring.

Use cases

  • Watching a small cluster of DGX Spark machines from one browser tab.
  • Tracking GPU temperature and LLM throughput on a home or lab workstation with an NVIDIA GPU.
  • Showing fleet status on a wall-mounted display using the kiosk page.
  • Feeding GPU and LLM metrics into an existing Prometheus or Grafana setup.

Things to know

  • Built mainly for NVIDIA DGX Spark; other units need SSH access and, for GPU data, nvidia-smi or Windows OpenSSH.
  • Several features, such as LLM metrics and the decode benchmark, depend on specific inference servers being reachable.

In collections

Self-hosted uptime and server monitoringSelf-hosted appsDashboards & business intelligence tools

Topics

#nvidia#gpu-monitoring#dashboard#llm#monitoring#ssh#prometheus#self-hosted

FAQ

What is sparkDash?
A self-hosted real-time web dashboard that monitors one or more NVIDIA DGX Spark and other GPU machines (GPU, memory, CPU, network, local LLM servers) in one browser window. sparkDash is a web dashboard for watching several NVIDIA DGX Spark (GB10) machines at once. It streams GPU, CPU, unified memory, storage, network and local LLM metrics over WebSocket, and units can be added, edited, reordered or removed from the UI without restarting the server. Linux machines with a discrete NVIDIA GPU can be added as dedicated GPU hosts over SSH, with system RAM and VRAM shown separately. Windows PCs with NVIDIA or AMD iGPUs and macOS machines are also supported as monitored units.
Is sparkDash free to use? What license does it have?
sparkDash is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is sparkDash still maintained?
Actively maintained. The last commit to sparkDash was on Nov 6, 2026.
What can I use sparkDash for?
Watching a small cluster of DGX Spark machines from one browser tab. Tracking GPU temperature and LLM throughput on a home or lab workstation with an NVIDIA GPU. Showing fleet status on a wall-mounted display using the kiosk page. Feeding GPU and LLM metrics into an existing Prometheus or Grafana setup.
Where can I get sparkDash?
The source code and installation instructions are on GitHub: https://github.com/MiaAI-Lab/sparkDash

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects