kinoshita-fj wrote: Thank you for your comments. I agree that we should not revert it. However, is this lowering beneficial for anything other than the use of pairwise load/store for 128-bit SVE? As I commented before, my understanding is that the practical benefit of this lowering itself is small. So, if the benefit is limited, considering the effect on other CPUs, wouldn't option 1 be better?
https://github.com/llvm/llvm-project/pull/127837 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
