[llvm] [NVPTX] Stop using 16-bit CAS instructions from PTX (PR #120220)

[Artem Belevich via llvm-commits]

Artem-B wrote:

> So, for this common use case performing the emulation direclty in LLVM by using CAS 32-bit results in better codegen than generating these CAS loops using atom.cas.b16.

Fascinating. Thank you for the detailed explanation.

https://github.com/llvm/llvm-project/pull/120220