jhuber6 wrote:

> We could possibly carve out an exception for `static inline` functions... but 
> I don't think that covers the cases that are failing. And it's not what gcc 
> does.

The fact that this required suppressing warnings globally in libc and libc++, 
which both have very intentional usage of these vectors, is very a big red flag 
to me. This seems to just turn on a "You used a wide vector!" warning on for 
everyone, which isn't very helpful in my opinion.

For `libmvec` we intentionally upscale from a `f32x8` to a `f64x8` which will 
now always fire a warning if I understand correctly. The original intent of 
these warnings from what I understood was mixing ABIs accross TU boundaries, 
which is why we force all of these to be static.

I don't have the full context here, but clang vectors are much more powerful 
than GCC vectors so I don't think we necessarily need to be beholden to them.

https://github.com/llvm/llvm-project/pull/199091
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to