andykaylor wrote:

> also, since it's a generic builtin (not X86-specific), I assume it should be 
> implemented in CIRGenBuiltin.cpp rather than CIRGenBuiltinX86.cpp. Right?

Yes. It seems that `BI__builtin_ia32_sqrtph512`, `BI__builtin_ia32_sqrtps512`, 
and `BI__builtin_ia32_sqrtpd512`. Were left in place by 
https://github.com/llvm/llvm-project/pull/165682, so we'll still need to handle 
them in `CIRGenBuiltinX86.cpp` (unless there is a plan to remove them), but the 
elementwise sqrt handling (along with standard sqrt builtin handling) should be 
done in `CIRGenBuiltin.cpp`.

https://github.com/llvm/llvm-project/pull/169310
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to