Abstract: A distributed memory management system which provides a novel usage of computer resources. Unlike memory management in conventional computers, memory management in the present invention is distributed, to avoid having data transfer requests wait in line for a centralized memory controller. Individual processes partition logical memory into layers of variable-sized subregions, then allocate these subregions for private use by individual processes. Within each partition, a process may use memory without concern for the activities of other processes. The regions are partitioned into sub-regions, and always include a central subregion C, and four peripheral subregions denoted west (W), south (S), east (E) and north (N). Only the central subregions are ever written to directly from outside the memory. Peripheral regions must first be partitioned before data can be written into their respective central regions.