A fast compacting memory management system for interactive systems

To write and benchmark a special-purpose allocator that can compact its heap, while still being fast enough for usage within non-trivial applications.

Main interface:
handle_t alloc(size_t);
void *lock(handle_t);
void unlock(handle_t);
void free(handle_t);

Compacting is done automatically or triggered by client code, e.g. when waiting for I/O.  The hypothesis is that by performing compacting when less costly, the operations free() and malloc() can be made O(1).

Inga kommentarer:

Skicka en kommentar