FORMAL MODEL OF DYNAMIC PARALLEL COMPUTATIONS IN PARALLEL COMPUTERS FOR EXPERIMENTAL DATA PROCESSING

Yu. E. Sheynin

St. Petersburg State University of Aerospace Instrumentation

Mass-parallel computers and distributed computer clusters are widely used in DAQ and off-line processing of experimental data as well as in complex instruments control and monitoring. Some of their tasks are dynamic in nature. A formal model of parallel computations is considered in correlation with a virtual machine features and its language, defining rules of operation for the uninterpreted (or partially interpreted) program schema. A model of parallel computations - Asynchronous Growing Processes (AGP-model) is proposed for parallel and distributed computing. It generates dynamic computations as self-modified dynamic networks of processes and controls them in asynchronous fully distributed manner.