arsenm wrote: I've switched this to leaving most of the useFP16ConversionIntrinsics usage in place. It should be removed, but it's been misused as an ABI option and I'm not sure how to untangle it. In particular I think some of the AArch64 builtins should be defined using _Float16 instead of __fp16
https://github.com/llvm/llvm-project/pull/174494 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
