Enabling your GPU to speed up when self-hosting on your computer

Hello to All, I just wanted to share my experience. Over the last few days I had tried several different setups. I am running Fedora 41 I have AMD Ryzen 9 7900 with 24 cores, 96GB Ram and a AMD Radeon RX 7600

Now I tried running n8n with docker desktop and then without docker using npm. When just running the demo workflow with n8n and I would bring up the chat and type in my question it would literary take 10+ mins for the LLM to spit something back to me. It was very frustrating.

Then I realized the Ollama Chat Model was just using my CPU, once I figured out how to enable the ROCm drivers the Ollama responded to my prompt in 2 seconds.

So if you are self-hosting on your our equipment and have the supporting hardware, enable the GPU for your docker images and n8n to use.

0 comments