Evaluation of the efficiency of automated system based on processors with VLIW and X86-64 architectures through testing

Abstract

The typical architectural concepts of popular processor families, their features, advantages and disadvantages are considered. Examples of structures of known processors are given. Technical limitations for the further development of popular RISC (CISC) architectures in accordance with the von Neumann machine paradigm are outlined. Attention is drawn to the increase in the size of the hardware scheduler and the resulting shortage of space on the chip with an increase in the number of functional modules on cores with a “sequential” structure. The advantages of the VLIW concept and the compiler working with this architecture are described. To assess the effectiveness of the functioning of automated systems with various architectures, a number of synthetic tests were carried out and test programs were used. Performance measurements were carried out to compare electronic computers based on VLIW and x86-64 cores. A number of algorithms have been proposed to accelerate computational tasks of an automated system with a parallel structure. The need to apply the concept of parallel programming when refining existing and developing new software for modern automated systems is emphasized. An example of converting program code into a parallel structure is given to describe an algorithm for a sequential task, illustrating complex information connections.