phoebewang wrote: > Change the definitions of `_mm_setr_ph`, `_mm256_setr_ph` and > `_mm512_setr_ph` to be functions instead of macros. Resolves #156709 > > Pending questions : > > * Should the `_mm_setr_ph` and `_mm256_setr_ph` functions be marked as > `__DEFAULT_FN_A TTRS128_CONSTEXPR` and `__DEFAULT_FN_ATTRS256_CONSTEXPR` > respectively ? I marked `_mm512_setr_ph` as > `__DEFAULT_FN_ATTRS512_CONSTEXPR` because this attribute list is also applied > to `_mm512_set_ph`. This is not the case for `_mm_set_ph` and `_mm256_set_ph`. #152910 only solves `_mm512_setr_ph`. So leaving `_mm_setr_ph` and `_mm256_setr_ph` as is would be fine.
> * What about the other `setr` intrinsics such as `_mm512_setr_ps`, > `_mm512_setr_pd`, `_mm512_setr_epi64`, etc... Should they be made into real > functions too ? If so, should it be done in this PR ? Yes, you can do it in the same PR. https://github.com/llvm/llvm-project/pull/156819 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
