arsenm added a comment. In D79744#2035283 <https://reviews.llvm.org/D79744#2035283>, @rjmccall wrote:
> > A completely different approach: OpenMP has to solve some very similar > problems and just lowers them completely in the frontend; have you considered > just doing that? Kernels need a ton of special-case handling anyway, and > IIUC you can never optimize over the boundary anyway. Yes, this is what I was describing. The problem is this ends up hurting optimizations because now we end up losing things like noalias on the pointer arguments. I also would still need to track the argument size/offset/align information in the IR, but for that we could keep on doing what we're doing and just never use the kernel arguments. One of the advantages of byval was an explicit align field to track this CHANGES SINCE LAST ACTION https://reviews.llvm.org/D79744/new/ https://reviews.llvm.org/D79744 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
