Would there be a hit? C++ has to do this: object pointer -> load vtable -> load function pointer while a compressed fat pointer would be like: fat pointer -

Would there be a hit? C++ has to do this: object pointer -> load vtable -> load function pointer while a compressed fat pointer would be like: fat pointer -> shift to extract vtable index -> shift to make into a global table offset -> load global table base from a global variable -> load function pointer Two loads in both cases, but in the compressed case the first one will almost always be cached; more code bloat, but not that much, and you save a bit of memory in the objects themselves.