Методики оценки производительности гетерогенных вычислительных систем Article is devoted to the performance evaluation of heterogeneous computing systems (platforms). The article substantiates the relevance of the use of problem-oriented configuration of computational systems for automated systems of information processing and management. The proposed mechanism is parallel and pipeline processing allowing to parallelize the task and thereby improve system performance. For interaction between modules it was proposed to use MPI-message passing interface between processes that perform a single task. The performance measurement was performed using recognized test kits Intel "MPI Benchmark" and "LINPACK HPL", which allow to fully assess the effectiveness of the computing system and its relevance for the user. Given the graphics performance of the heterogeneous computing complex "profit center" in dealing with complex problems of linear algebra. The results obtained confirm high efficiency of MVP for various applications.

The article describes the developed computational complex of morphological image analysis for interferometric side-scan sonars. Examples of modeling the operation of the side-scan sonar in cases where the structure of bottom sediments is not known in advance are given. The results of the analysis of experimental data are considered. A description is given of the structure for constructing algorithms for morphological image analysis used in the operation of the complex. Recommendations are given on the effective use of the computing capabilities of the Griffon hardware platform for organizing parallel-conveyor data processing of side-scan interference sonars in real time. Hardware and software modeling has been performed to assess the most important performance characteristics of a computing complex and determine the optimal organization of data processing. The obtained results of processing sonar images streams showed that the use of the Griffon hardware platform provides advantages in terms of organizing the computing process, including parallel operation of different sections of the computing pipeline, reducing the load on the transport bus and pipeline delay, as well as reducing the load and freeing up CPU resources for morphological processing.

The article contains materials on methods for designing protected high-performance heterogeneous computing platforms (VGVP) based on the modular architecture of the CompactPCI Serial standard. A review of the development of Intel’s x86 processor technology, a review of common video interfaces is given. The requirements for the projected heterogeneous hardware-computing platforms for solving video processing tasks are determined. Platforms that organize parallel processing of digital video images using GPU modules, FPGA modules and CPU modules are described. It is shown that the implementation of linear filtering algorithms on FPGAs allows processing images up to 10 times faster than using GPU modules and even more CPU modules. The maximum power consumption and the capacity of the VGVP with conductive cooling are estimated.

В статье обоснована разработка научного подхода проектирования высокопроизводительной гетерогенной вычислительной платформы (ВГВП) для обработки изображений. В качестве прикладной проблемы для реализации на ВГВП выбрана задача обработки изображений в режиме жесткого реального времени, поступающей по каналам связи от разнородных специализированных информационных систем. Рассмотренные научные подходы реализуют концепцию создания многомодульной архитектуры, в которой одновременно несколько вычислительных модулей с разнородной архитектурой могут параллельно обрабатывать информационные потоки. В статье показано, что гетерогенность, архитектурные решения и программные механизмы взаимодействия модулей различной архитектуры позволяют эффективно применять ВГВП для гетерогенной параллельно-конвейерной обработки данных.

The ways of improving and the principles of constructing specialized computers that provide control and signal processing in mobile and stationary over‑the‑horizon radar systems are discussed in the article. A brief review of the history of the development of over‑the‑horizon radar is presented. Potential capabilities of the equipment are considered, demonstrating the prospects of creating specialized computers based on the Gryphon hardware platform with high parameters and minimal mass‑size characteristics. The structure and principles of constructing the mathematical support of specialized computers are described in details. Algorithms are proposed that provide an increase in the efficiency of signal processing, taking into account the structure of electromagnetic fields formed during the propagation of waves in the ionosphere. The comparative characteristics of the proposed algorithms that can be used to optimize signal processing and digital equipment volumes are given.