This Report was presented to the Faculty of the Graduate School of the University of Texas at Austin in partial fulfillment of the requirements for the degree of

Master of Science in Engineering

This report outlines some of the approaches taken to analyze the performance and capabilities, or benchmark, embedded microprocessors and microcontrollers. Beginning with the advent of desktop computer benchmark suites, as both an engineering evaluation tool and a marketing tool, the use of benchmarks as a measuring stick for processor-based system performance has become pervasive. While many of the popular benchmarks used to measure the performance of desktop or laptop computers provide a reasonable system evaluation in that environment, the attempts to apply similar benchmark suites to embedded systems has so far met with little success. This is primarily due to the large control aspect of many embedded systems, in which hardware and software work together to produce the desired outcome. Because traditional computer benchmarks are based primarily on the speed at which data is processed, these benchmarks are rarely relevant to embedded applications. Some industry efforts are currently underway to develop benchmark suites which are more appropriate for measuring the performance of an embedded system, but the diverse and application-specific nature of most embedded systems make this a daunting task. This paper will review some of the previous and ongoing approaches used for benchmarking the performance of microprocessors and microcontrollers in embedded systems, and discuss why these have met with limited success. This paper proposes a comprehensive approach which can be taken to benchmark embedded microprocessors and microcontrollers, including two application examples to illustrate the difficulties which must be overcome when developing a valid benchmark suite for the embedded applications space.