Journal "Software Engineering"
a journal on theoretical and applied science and technology
ISSN 2220-3397

Issue N9 2015 year

Dynamics Control Computing in the Processor Data Flow Architecture for Different Types of Algorithms
V. M. Bakanov, Professor, e-mail: vbakanov@hse.ru, National Research University Higher School of Economics, Moscow

The paper set out the goals of definition of rational modes of calculators post Von Neumann architecture with automatic parallelization of hardware (streaming, DATA FLOW architecture) and proposes solutions to some of them. Evaluators streaming architecture have undoubted advantage in the form of a fully automatic parallelization of data processing at the hardware level and can be considered as an alternative future of modern processors classical von Neumann architecture. In contrast to the traditional architecture of calculators in this case, you can control the intensity of computation (number of simultaneous operations), defined the conditions for such management. Targeted control of the intensity calculation is useful in terms of rational use of both existing performing devices and loads (traffic) inside chip data bus and allows more efficient use of calculators streaming in single and multi-tasking. Management strategies implemented by setting priorities sample ready to run operators from the buffer memory streaming command calculator, with the possible intensification of regimes as well as the intensity of depression calculations. The concept of function computing complexity of expanding its change in time of the program. These tasks performed by computer simulation shows the multiplicity of solutions and the ability to choose the most effective one. The effectiveness of several proposed strategies for managing the intensity calculation is shown by the example of several commonly used standard algorithms.

Keywords: parallelization of data processing problems paralleling paralleling automation, software and hardware parallelism, post the von-Neumann architecture, streaming calculators, of computing-intensive, computer simulation, strategy, controlling the intensity calculations
pp. 20–24