Wolfram70 wrote:

> Should we also move the FTZ and/or SAT knobs to immediate args?

I agree with this but currently, I think that will make the 
codegen/verification more complicated since the matrix of support for these 
with all the floating-point types is kind of sparse. Since we now have support 
for default arguments in intrinsics, we could move these to immediate 
default-valued arguments at any time if needed without breaking any existing 
usage as well.

https://github.com/llvm/llvm-project/pull/217336
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to