Nazarova I. Rise of the efficiency of the parallel methods to decision of dynamic tasks in multiprocessor calculable systems

The goal of the work: rise of the efficiency of multiprocessor computer systems. Research object: parallel computer systems with the distributed memory. Subject of research: parallel methods of decision of dynamic tasks with the concentrated parameters. Methods of researches: theory of ordinary differential equations, calculable mathematics, theory of the graphs and algorithms. Scientific novelty: the obvious parallel methods of decision of non-stiff tasks on the basis of extrapolation of Richardson are improved; the parallel block embedded and extrapolation methods of decision of stiff tasks are first offered; the parallel exponential methods of integration of linear system ODE's with control of local error are first offered; the block parallel algorithm of increase of dense matrices is improved. Introduction: "Scientific technical centre of saving energy problem", educational process of departments of PMI DonNTU and technical informatics of State universitety of informatics and artificial intelligence. Sphere of the use: computer systems and networks with parallel architecture .