For a similar project I was on -- that is a managed language built on top of LLVM -- we ensured GC events only came from calls to GC functions, either explicit

For a similar project I was on -- that is a managed language built on top of LLVM -- we ensured GC events only came from calls to GC functions, either explicit collects or allocates. Spills to the stack of references only occurred in calls to functions and not loops. Our collector was an accurate collector though. This project was targeting game code for the register heavy, branch-unfriendly, in-order PowerPC PS3 and Xbox 360. Performance in scalar numerical code was equal to equivalent C++ code.