System Benchmark

Hardware performance scoring with NOMAD Score, Builder Tags, and a community leaderboard.

The System Benchmark tests your hardware and produces a NOMAD Score — a weighted composite of CPU, memory, disk, and AI performance. You can compare your results against other N.O.M.A.D. builds on the community leaderboard at benchmark.projectnomad.us.

How to run a benchmark

1. Open the System Benchmark

   Go to Settings → System Benchmark, or navigate to /settings/benchmark.

2. Choose a benchmark type

   Select one of the three available types (see Benchmark types below).

3. Run the benchmark

   Click Run. The benchmark runs in the background and may take several minutes to complete. Do not shut down or restart your server while it is running.

4. View your results

   When the benchmark finishes, your NOMAD Score and individual component scores are displayed.

Benchmark types

Type	What it measures	Can be shared to leaderboard
Full	CPU, memory, disk, and AI performance	Yes — if AI Assistant is installed
System Only	CPU, memory, and disk only	No
AI Only	AI inference performance only	No

Only Full Benchmarks that include AI results can be submitted to the community leaderboard.

NOMAD Score

The NOMAD Score is a single weighted composite number calculated from your benchmark results across four categories:

• CPU performance
• Memory performance
• Disk performance
• AI inference performance

The AI component contributes meaningfully to the overall score, so systems with GPU acceleration will generally score higher than CPU-only builds.

Builder Tags

A Builder Tag is a NOMAD-themed identity you assign to your benchmark result before sharing it to the leaderboard. Tags follow the format Word-Word-0000 — for example, Tactical-Llama-1234.

Builder Tags are optional for local results but are associated with any submission you share publicly.

Sharing results to the leaderboard

You can submit qualifying results to the community leaderboard at benchmark.projectnomad.us.

Requirements to share a result:

• The benchmark must be a Full Benchmark (not System Only or AI Only)
• The AI Assistant must be installed and working — the benchmark must include AI results
• Your score must be higher than any previous submission from the same hardware

If submission fails, check the error message. The most common reasons are: the benchmark type was not Full, the AI Assistant is not installed, or the new score is not higher than the previous submission from that build.

Measuring GPU impact

The System Benchmark is a practical way to quantify the effect of hardware changes on your N.O.M.A.D. build.

Run a Full Benchmark before adding a GPU, then run another after installing the GPU and completing the AI Assistant Force Reinstall. The difference in NOMAD Score reflects the real performance improvement — GPU-accelerated builds typically see AI speeds of 100+ tokens per second compared to 10–15 on CPU only.

See the AI Assistant page for instructions on enabling GPU acceleration.

Related

AI Assistant

Install the AI Assistant and enable GPU acceleration to maximize your NOMAD Score.

System health

Monitor CPU, memory, and storage usage on your server.