Realization Features of System Software of Multiprocessor Computing Systems

Realization Features of System Software of Multiprocessor Computing Systems

Boris Moroz (University of Customs and Finance, Ukraine), Eugene Fedorov (Donetsk (Pokrovsk) National Technical University, Ukraine), Ivan Pobochii (National Metallurgical Academy of Ukraine, Ukraine), Dmytro Kozenkov (National Metallurgical Academy of Ukraine, Ukraine) and Larisa Sushko (Dnipro State Agrarian and Economic University, Ukraine)
Copyright: © 2019 |Pages: 21
DOI: 10.4018/978-1-5225-7588-7.ch016

Abstract

The chapter is aimed at the problem of use and adjustment of system software of multiprocessor computing systems. The main principles of the Linux operating system were considered, which were necessary when constructing a multiprocessor computing system. These studies also cover new ways to remotely access the memory of processor systems through the use of RDMA technology for InfiniBand technology. Thus, it has been shown that the RDMA principle, together with the formation of a separate computing network in the data interchange environment, and the implementation of VLAN mechanisms, allowed the data transmission among nodes memory of the multiprocessor computing system without additional buffering. This approach does not require the active OS operation, libraries, or applications on those nodes of the system which memory is requested.
Chapter Preview
Top

Background

Nowadays there is a rapid increase in the number of multiprocessor computing systems and their total productivity in the world (Bashkov et al., 2013; Shvachych et al., 2008; Latsys, 2003; Voevodyn & Zhumatyy, 2007). This is due to the fact that such systems became publicly available and cheap hardware platforms for high-performance computing (Ivaschenko et al., 2017; Shvachych et al., 2018; Ivaschenko et al., 2018; Shvachych et al., 2018). At the same time, the interest has sharply increased in the problems of the computer networks (GRID) and awareness of the fact that their implementation will have a huge impact on the development of human society, is comparable to that of the emergence of unified electric networks at the beginning of XX century. In this regard, considering the problems of mastering multiprocessor computing systems, it should be taken into account that they are the first step in the creation of such computing networks.

The more so the practice puts a variety of problems before the application scientists, the full solution of which in most cases is possible only by multiprocessor computing systems (Alishov et al., 2016; Ivaschenko et al., 2005). Obviously, today, based on multiprocessor information and management systems, control systems are being implemented for many industries: the space industry, aviation, anti-aircraft and anti-missile defense systems, and many others. However, the production of multiprocessor information and management systems is hampered by the high costs at all its stages. As a result, the overall cost of a system often makes it an inaccessible tool. The use of modern multiprocessor cluster systems would reduce the cost of its production (Ivashchenko et al., 2011; Bashkov et al., 2011).

For the aviation industry, the development and implementation of new materials are relevant. And here an important component of the reliability of aircraft is the interaction with metallurgical production. At the same time, in the metallurgical industry, there is a lot of diverse and interconnected processes. First and foremost, these are the technologies of smelting and casting of iron-carbon alloys, heating, rolling and heat treatment of metal products (Shvachych et al., 2003; Ivaschenko et al., 2003), as well as the operation of auxiliary equipment, which includes filling machines, buckets, bowls, and so forth. Industrial practice shows that neither the intensification of the processes of metallurgical production nor the constructive improvement of various metallurgical equipment is possible without studying and analyzing the phenomena of heat and mass transfer (Ivaschenko et al., 2005; Shvachych et al., 2007). At the same time solving these problems by well-known standard approaches is a complex problem, the overcoming of which is possible only through the use of modern multiprocessor computing technologies. At the same time, one of the main features of the introduction of such technologies is to increase the speed and computing productivity (Shvachych et al., 2018; Shvachych et al., 2017). High-performance computing allows to solve multi-dimensional problems, including those that require a large amount of processor time (Ivaschenko et al., 2018; Shvachych et al., 2018). Speed ​​control makes allows to efficiently manage technological processes, or even create the preconditions for developing new promising technological processes.

Key Terms in this Chapter

Cluster Computing Systems: Multi-processor computing systems based on local networks, known as “cluster computing systems.”

TFTP Protocol: Trivial file transfer protocol is a simple file transfer protocol. It is mainly used for initial downloading of diskless slave nodes.

Supernumerary Devices: In fact, this is about multiprocessor parallel computing systems of MPP architecture (massively parallel processing).

Communication Environment: For making supercomputers there are used serial microprocessors, each with its own local memory and connected via some communication environment.

InfiniBand Technology: The data interchange among computing nodes of a multiprocessor computing system is transferred to a separate network using the InfiniBand network interface, which increased the system's overall performance and significantly reduced the latency (channel download) of the network that connects nodes of the cluster computing system.

Blade-Server Solutions: In this chapter, the authors consider multiprocessor computing systems with several similar motherboards installed in the same package. Practice shows that “blade” systems are more compact and easy to maintain, and their implementation is not much more expensive compared to multiprocessor computing systems.

DNS Server: Domain name system server (bind-9.2.4-27.0.1.el4) of the main module performs direct and inverse transformation of node names into IP addresses.

Vectorization of Computations: The creation of parallel computing systems also required the development of a new approach for the making appropriate computing algorithms. For this reason, new ideas and approaches in the field of computational mathematics are now emerging, which are aimed towards the development of new methods for the implementation of numerical experiments based on the computations’ parallelization.

Complete Chapter List

Search this Book:
Reset