Benchmarking ARM-based Application Integrated Systems

Abstract

ARM Holding Ltd. is in a very good position; by owning a very prolific instruction set architecture (named ARM) and licensing it to other companies that compete with one another, they are always winners. Though ARM is used throughout the domain covered by embedded systems; one particular market that they have cornered is in the application cores used in mobile handset devices. Licensees oftentimes make their own alterations to the core, add other system components, and print or package them together into devices that this thesis will refer to as integrated systems. Though finding very specific architectural details between the cores themselves is as simple as going onto the website, true performance comparisons stemming from credible benchmarks of the integrated systems that contain them are not as forthcoming. This thesis aims to explore the realities of benchmarking ARM-based application integrated systems. By looking at what kind of benchmarks are available and what kind of tests have been done in the past it should become apparent what, if anything, needs to be developed to provide thorough comparisons between these devices. During the course of this investigation, topics such as the ambiguity in integrated system methods, the future of application integrated systems, and the process in actually selecting and running a benchmark on a sample system shall be explored.