wentbackward

wentbackward / nv-monitor

Public

A lightweight terminal system monitor built for the **NVIDIA DGX Spark** (Grace CPU + GB10 GPU).

94
13
100% credibility
Found Apr 01, 2026 at 94 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
C
AI Summary

A compact terminal program providing real-time visual monitoring of CPU, memory, and NVIDIA GPU metrics, with data logging and export options.

How It Works

1
🔍 Discover nv-monitor

You hear about a handy tool that watches your computer's processor, memory, and graphics card in one colorful screen.

2
📥 Get the program

Download the tiny file to your Linux computer and get it ready with a quick setup.

3
🚀 Launch the dashboard

Start the program to see live bars and charts of CPU usage per core, memory levels, GPU heat, power use, and running processes updating smoothly.

4
💾 Record your stats

Turn on saving to a simple data file or share numbers with tracking tools for later review.

5
⚡ Test with activity generator

Run the fun load creator to make wavy patterns of busyness and watch everything respond perfectly.

✅ Effortless monitoring

Sit back as your NVIDIA system stays perfectly watched with accurate, low-overhead insights anytime.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 94 to 94 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is nv-monitor?

nv-monitor is a lightweight terminal monitor for Linux systems with NVIDIA GPUs, delivering a ncurses-based TUI that tracks CPU per-core usage (with ARM big.LITTLE labels like X925/X725), memory (fixing HugePages inaccuracies on DGX Spark), GPU utilization, power, temperature, VRAM or unified memory, encoder/decoder stats, and top GPU processes. Built in C as a single <80KB binary with zero runtime dependencies beyond ncurses and NVML, it also offers CSV logging (`./nv-monitor -l stats.csv`), headless mode, adjustable refresh (`-r 2000`), and a Prometheus exporter on a port (`-p 9101`). A bundled demo-load tool generates synthetic CPU/GPU loads for pipeline testing.

Why is it gaining traction?

Unlike generic monitor nv failsafe tools or bloated alternatives, it nails DGX Spark quirks like unified memory and HugePages without overhead, packing TUI charts, process sorting (toggle with 's'), and OpenMetrics export into one binary—ideal for clusters where lightweight terminal for linux beats heavy clients. Developers grab it from github lightweight github for its no-build hassle (prebuilt binaries), dynamic NVML loading (works sans full NVIDIA deps), and low-footprint Prometheus scraping, standing out amid github lightweight charts python or vk options.

Who should use this?

GPU cluster admins on DGX Spark or x86 NVIDIA Linux rigs needing real-time TUI oversight during inference runs. AI/ML engineers validating pipelines with demo-load, or ops folks piping metrics to Prometheus for g4s secure monitoring nv without agent sprawl. Skip if you're on macos/windows—it's Linux-only, like lightweight terminal emulator linux.

Verdict

Grab it for NVIDIA Linux monitoring if you want a no-fuss, accurate TUI/exporter; 87 stars and 1.0% credibility score signal early maturity, but MIT license, detailed docs, and binaries make it low-risk to test. Pairs well as lightweight github alternative to htop+nvidia-smi stacks.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.