Date of Award

Spring 5-23-2017

Document Type

Undergraduate Honors Thesis


Mechanical Engineering


Frank Jacobitz


David Mayhew


In the realm of scientific computing, it has become increasingly important to focus on results driven growth (Kamil, Shalf and Storhmaier). Doing this enables researchers to continue building the rapidly expanding area of scientific discovery. However, with the growth comes a cost of the amount of resources consumed to accomplish these results. Large supercomputers are consuming power at a rate roughly fourteen thousand times that of a traditional American household (U.S. Energy Information Administration). Parallel to this, public consumers have been driving the mobile industry and the research behind it. The need to have faster and faster mobile devices that can last all day long on a single battery charge has driven the development of Advanced Reduced Instruction Set processors developed by ARM. These processors are built to perform efficiently while still maintaining the ability to perform the necessary calculations. This study looked at combining these two parallel realms and analyzing the overall efficiency and energy consumption of multiple ARM processors as compared to a traditional desktop computer. The results showed that the ARM processors were less efficient roughly by an order of two when compared to the slowest possible trial on the desktop. Several variables played a significant role in these results including the limitation on network speed and bandwidth, idle energy consumption, and individual power regulators.