A method for compaction of objects within a computer memory, the method including dividing a memory space into a plurality of non-overlapping sections, selecting a plurality of source sections from among the sections, each containing at least one object, selecting a plurality of target sections from among the sections, and moving any of the objects from the source section to the target section, where each of a plurality of pairs of the source and target sections is exclusively available to a different process from among a plurality of processes operative to perform any of the steps with a predefined degree of concurrency.