sepavloff added a comment. In D153339#4489025 <https://reviews.llvm.org/D153339#4489025>, @aaron.ballman wrote:
> This seems reasonable to me, but I do wonder if changing the return type from > int to bool will cause surprises for anyone. I see why it's done, but this > seems more like a C interface (to me) which would return an `int`; returning > a `bool` gets promoted to `int` when you sneeze near it, so I wonder about > performance implications. But at the same time, we wouldn't want a vector of > ints returns in the vector case, I would suspect. The new implementation uses the representation close to OpenCL, when it makes result of comparisons: https://registry.khronos.org/OpenCL/specs/3.0-unified/html/OpenCL_C.html#operators-relational. It is a vector of integers, where each element is of the same bit size as the source vector element. They must be already supported by the codegen. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D153339/new/ https://reviews.llvm.org/D153339 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
