One embodiment of the invention provides a method and apparatus for decompressing a compressed data set using a processing device having a plurality of processing units and a tangible, non-transitory shared memory. The compressed data set comprises a plurality of compressed data segments, in which each compressed data segment corresponds to a predetermined size of uncompressed data. The method includes loading the compressed data set into the shared memory so that each compressed data segment is stored into a respective memory region of the shared memory. The respective memory region has a size equal to the predetermined size of the corresponding uncompressed data segment. The method further includes decompressing the compressed data segments with the processing units; and storing each decompressed data segment back to its respective memory region within the shared memory.