Hi,
If you're looking for a simple and fast allocator, may I recommend
bztalloc, which has several advantages over dlmalloc, ptmalloc, Hoard and SLAB:
* minimal dependencies (memcpy, memzero, mmap, munmap, that's all, and mmap is always called anonymously without an fd)
* it does not mix allocator data and application data
* can be used in single threaded and multi threaded environments
* can support any threading library, not just pthreads
* works on so called arenas, so you can use the same library for kernel heap, user space, shared memory etc. at the same time
It worth mentioning that jemalloc() usually outperforms it, but jemalloc() is a lot more complex and lot more difficult to port. This bztmalloc is a nice compromise between performance, low memory footprint and easy portability.
Cheers,
bzt