================ ---------------- phoebewang wrote:
When AVX512BF16 is enabled, it's lowered to `vcvtneps2bf16`, https://godbolt.org/z/d5snYdGn4. Otherwise, it's scalarized with libcall. Since we change for intrinsic, AVX512BF16 is guaranteed. I don't see any risk here. https://github.com/llvm/llvm-project/pull/169841 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
