This reduces our codegen size and removes a few umov instructions. Performance falls within noise but this small change will allow us to do more vector optimizations in C code in the future.