YonahGoldberg wrote:

> If `elementwise` is not specified, and if the vector is smaller than the 
> native size supported by the backend, and the backends don't override them, 
> then AtomicExpand should just expand these with a cmpxchg loop. Otherwise, we 
> don't have builtins for these, so the backend should just fail.

Sounds good, I think this matches my current implementation.

> If `elementwise` is specified, we can fragment the atomic into smaller 
> atomics, like we do for FP vectors.

Well elementwise expansion will be left to a my other PR, but I think the 
elementwise expansion shouldn't be the default. We should by default treat 
elementwise atomics as normal atomics.



https://github.com/llvm/llvm-project/pull/190716
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to