3 ms·
Yes. We have a quad MI300A server and run several inference models on it. For $107k it has saved us so much money on tokens already and it's a heck of a lot fas
by FuriouslyAdrift 3mo ago
Yes. We have a quad MI300A server and run several inference models on it. For $107k it has saved us so much money on tokens already and it's a heck of a lot faster than cloud services.
- anon291 3mo agoDo you do research? Because if you don't, then you're not the target market for general purpose AI compute chips.
- FuriouslyAdrift 3mo agoWe use it for business analysis in a manufacturing environment (inputs from sensors, databases, external signals, etc.) We also use it as a programming tool to design software, ingest logs and do analysis, find patterns in data, etc. We bought it because we were burning through too much cash with Claude and Gemini.