>>26
Seriously. Writing a small object allocator in C++ (by overloading operator new) which allocates a large chunk of memory once then "allocates" further instances from this chunk is a pretty simple thing.
It's a well known "flaw" that C's dynamic memory allocation system (and thus C++'s as well - most implementations have an operator new which just wrappers malloc) is designed for allocating large blocks of sequential memory. Allocating small objects on the heap is slow as fuck, and implementing a small object allocator (or slab allocator, or whatever) is a great way to get a lot of performance.
But I guess you wouldn't understand most of that, coming from a garbage-collected fagfest. Go back to fapping at your forced OO paradigm while I have sex with my CFLAGS.