This is a series of articles looking at the Minisforum AI X1 Pro running Linux. In this series, I examine every aspect of this Mini PC in detail from a Linux perspective.
The Minisforum AI X1 Pro is a powerful mini PC based on an AMD Ryzen AI 9 HX 370 with integrated Radeon 890M graphics. This is no ordinary processor and no ordinary integrated GPU. The machine can handle extremely demanding tasks. With 64GB of RAM and 1TB NVMe this configuration retails for £929.
In my benchmarks article, the Minisforum’s Radeon 890M scores 11970 running glmark2. That score is considerably higher than an Intel Arc 140T’s score of 6914. I’m always wary of this type of benchmark and how it translates to the performance of AI programs. A reader emailed me asking to perform a benchmark with an actual AI program. I’m happy to oblige.
When it comes to AI software, NVIDIA has the upper hand, with a lot more open source software supporting NVIDIA cards than AMD or Intel GPUs. Looking for a quick test, Jan sprung to mind. Jan is a ChatGPT alternative that runs 100% offline on a Linux machine. The aim of the project is to make it easy for anyone to run Large Languages Models and use AI yet retain full control and privacy.
Many AI programs are non-trivial to get running particularly when you want to run a specific program on NVIDIA, AMD, and Intel GPUs. Jan is probably the easiest AI program out there for this purpose and its cross-platform AppImage doesn’t need any configuration to get the software working with the 3 different GPUs.
I asked Jan a simple question: What is LinuxLinks.com
With the Minisforum AI X1 Pro with its Radeon 890M graphics it generates 25 tokens/sec with the Jan-V1-4B-GGUF model. Jan-v1 is the first release in the Jan Family, designed for agentic reasoning and problem-solving within the Jan App.
“Tokens/sec” is a metric used in AI and API services to measure how many language model tokens an API can process in one second.
Using the same question and model, the Arc 140T is much slower at generating a response. But both the iGPUs are eclipsed by a mid-range dedicated NVIDIA graphics card.
Whatever graphics card you’re using, try out the same test on your machine using the Jan-V1-4B-GGUF model.
In the next article in this series I’m going to look at an AI program that’s harder to install.
Complete list of articles in this series:
| Minisforum AI X1 Pro | |
|---|---|
| Introduction | Introduction to the series and interrogation of the Minisforum AI X1 Pro |
| Benchmarks | Benchmarking the Minisforum AI X1 Pro |
| Power | Testing and comparing the power consumption under various workloads |
| Jan | ChatGPT without privacy concerns |