because malloc is too slow
- slab allocator optimized for rapid alloc/dealloc of small objects
- memory managed in fixed sized blocks
- blocks cached per thread to limit concurrency overhead
- block distribution between threads handled in global pool
- 200+% faster than malloc
WIP Todo:
- block alignment
- mmap