[FuriouslyAdrift]

AMD Instinct is their direct competitor for compute and they are better per dollar, better per watt, and out competing on raw performance.

Only thing holding them back is fab capacity which nVidia keeps buying in bulk to keep them small.

[HDThoreaun]

AMD is held back by their interconnect and firmware disadvantage compared to nvidia. They’ve been trying really hard to create their own cuda, but rocM and HIP still aren’t very popular especially for research.

[pocksuppet]

And their repeated refusal to either implement CUDA or reimplement everyone's CUDA libraries on their own platform. They say that AMD never misses a chance to miss a chance.

[anon291]

Have you ever actually had anyone work with these chips? Developer ux on amd is terrible.

[FuriouslyAdrift]

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]

Do you do research? Because if you don't, then you're not the target market for general purpose AI compute chips.

[FuriouslyAdrift]

We 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.

[lelanthran]

> Have you ever actually had anyone work with these chips? Developer ux on amd is terrible.

Just how much of dev ux do you need? A foundational library, of course, but as the AI companies keep saying, their models can vibe-code what's needed for those chips anyway.