[llvm] [NVPTX] Prefer prmt.b32 over bfi.b32 (PR #110766)

[Artem Belevich via llvm-commits]

Artem-B wrote:

Indeed. According to https://arxiv.org/pdf/2208.11174 `BFI` is much more expensive than `PRMT` which appears to take just 1 cycle on A100:
![image](https://github.com/user-attachments/assets/577e75c7-d632-42f8-b642-963baef8103b)


https://github.com/llvm/llvm-project/pull/110766