COARSE-GRAIN PARALLEL COMPUTING FOR VERY LARGE-SCALE NEURAL SIMULATIONS IN THE NEXUS SIMULATION ENVIRONMENT

We describe a neural simulator designed for simulating very large scal e models of cortical architectures. This simulator, NEXUS, uses coarse -grain parallel computing by distributing computation and data onto mu ltiple conventional workstations connected via a local area network. C oarse-grain parallel computing offers natural advantages in simulating functionally segregated neural processes. We partition a complete mod el into modules with locally dense connections-a module may represent a cortical area, column, layer, or functional entity. Asynchronous dat a communications among workstations are established through the Networ k File System, which, together with the implicit modularity, decreases communications overhead, and increases overall performance. Coarse-gr ain parallelism also benefits from the standardization of conventional workstations and LAN, including portability between generations and v endors. (C) 1997 Elsevier Science Ltd.