[llvm] AMDGPU][True16][CodeGen] FP_Round f64 to f16 in true16 (PR #128911)

[Brox Chen via llvm-commits]

broxigarchen wrote:

> > I don't understand this change. You shouldn't have to make any of this code conditional on true16. Are you just avoiding the FP_TO_FP16? We can and should do that independently of true16
> 
> I thought that FP_TO_FP16 should only be used on targets that put fp16 values in 32-bit registers, which AMDGPU traditionally does but True16 does not.

Hi @arsenm and @jayfoad. Thanks for the review! Do you think it's good to continue with this implementation?

https://github.com/llvm/llvm-project/pull/128911