https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86141
--- Comment #18 from ASA <aaron_sami_abassi at hotmail dot com> --- Eventually this optimization problem can start to have severe implications. Consider the contrasting Assembly output of GNU C++ versus LLVM C++ in the following case: Original source code: https://github.com/ASA1976/RAP-BTL/blob/master/fastlist.cpp GNU Assembly output: https://github.com/ASA1976/RAP-BTL/blob/master/fastlist-g%2B%2B.s LLVM Assembly output: https://github.com/ASA1976/RAP-BTL/blob/master/fastlist-clang%2B%2B.s 299 lines of Assembly code which execute in a single stack frame versus 1711 which execute in many stack frames.
