I had ChatGPT walk me through a cluster configuration of 2 ASUS GX10 (DGX Spark) machines, to run a single model on the shared memory of both machines. Here's the full chat transcript:
- https://com-pute.com/nick/Clustering%20Asus%20GX10s.mhtml
- https://com-pute.com/nick/Clustering%20Asus%20GX10s.txt
(Download the mhtml file, and you can open it in Chrome or Edge)
It's actually quite a bit to set up the first time, but after that, configuring new LLMs is quick once you have the models downloaded.