zlacker

[return to "AMD funded a drop-in CUDA implementation built on ROCm: It's now open-source"]
1. 2OEH8e+Yi[view] [source] 2024-02-12 15:35:34
>>mfigui+(OP)
Question: Why aren't we using LLMs to translate programs to use ROCm?

Isn't translation one of the strengths of LLMs?

◧◩
2. JonChe+Cy[view] [source] 2024-02-12 16:40:29
>>2OEH8e+Yi
You can translate cuda to hip using a regex. LLM is rather overkill.
[go to top]