[clang] [llvm] [opt][AMDGPU] Add pass to handle AMDGCN pseudo-intrinsics target specific info), start with `llvm.amdgcn.wavefrontsize` (PR #114481)

[Joseph Huber via cfe-commits]

jhuber6 wrote:

> The toothpaste is out of the tube once the IR is produced. If some toolchain were relying on the global target machine features, there are opportunities for error on each tool invocation. The absence of the attribute does not tell you what the final compilation context will be.

The absence of the attribute just says can we can't statically assert that it's one value or the other, keeping it is the pessimistic route, but it should be correct AFAICT.

https://github.com/llvm/llvm-project/pull/114481