Hardware simulation in distributed computing systems: Methods and tools (Q1918738)

By a computing system (CS) we mean the operating of a single- or multi-processor computer, which consists of hardware, programs loaded and interpreted by this hardware, and data processed by the programs. The CS hardware consists of integrated circuits on a chip. The most advanced chips to day contain several millions logic components, and forecasts indicate that by the year 2000 chips with hundreds of millions of components will be common in CS. Already today CS software contains many millions of machine instructions. The essential increase in the integration scale of computer chips leads to an accelerating growth in the complexity of the space of design solutions for new integrated circuits on these chips and systems assembled from circuits. In other words, the complexity of the design problem for new CS is rapidly growing. The development of new CS is impossible without further development of computer-aided design technologies, which incorporate simulation methods and tools as an inseparable component. In simulation of a new CS, we first identify the block of problems that study the properties and the characteristics of the CS hardware. A central issue here is the description of the hardware in design languages. In the process we inevitably study also some of the CS software, but its volume is not large and it is determined primarily by the simulation hardware used. The software for new CS is usually studied using special languages and methods of system simulation, which are not considered in this article. In our view, the CS simulation arsenal should include tools for both hardware and software simulation, but the elaboration of this thesis is a subject for another article. In this article, we consider the requirements that must be met by algorithmic simulation and design languages for single- and multi-processor CS, present examples of algorithmic design languages, and examine methods of sequential and distributed simulation of CS hardware. The development and application of distributed simulation tools and methods is a highly topical problem, because in the foreseeable future simulation problems for new chips and CS will become too complex to be solvable even on the fastest among the existing single-processor computers by conventional sequential methods.