AMD Watch posted an update
AMD CTO Mark Papermaster says some scientific computing tasks need native FP64 precision, arguing that AI approximations alone are not enough for work such as fusion research, molecular studies and atmospheric modelling.
Why it mattersThat is a useful distinction as AI moves deeper into research: some workloads benefit from approximation, while others need high-precision computation. No, the latest model is not automatically the right tool for every equation. Papermaster made the comments in an interview with Scientific Computing World. For researchers, where should the line fall between AI-based approximation and traditional high-precision computing?
Discuss: For scientific workloads, where should researchers draw the line between AI-based approximation and native high-precision computing?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.