yxsamliu wrote: > It looks like this is introducing a new heuristic that is based on the alloca > type. @vtjnash is currently actively working on removing dependencies on the > alloca type, so I think we should avoid this. > > Can this information be derived based on access patterns instead? In the > motivating case, it looks like there are i64 stores on which this can be > based?
I think it might be feasible. Let me try that. https://github.com/llvm/llvm-project/pull/165159 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
