Nanoprocessors : Configurable Hardware Accelerators for Embedded Systems

Title of Thesis: NANORPOCESSORS: HARDWARE ACCELERATORS FOR EMBEDDED SYSTEMS Lei Zong, Master of Science, 2003 Thesis directed by: Professor Bruce L. Jacob Department of Electrical and Computer Engineering Today’s consumer market is driven by technology innovations. Many technologies that were not available a few years ago are quickly being adopted into common use. Equipment for these services requires microprocessors inside and can be regarded as embedded systems. Embedded systems are computer systems that are well hidden inside devices. At the time of design, much is known about the operating conditions and requirements. Embedded systems are designed to meet these requirements at a minimal cost. To improve efficiency and throughput, real-time operating systems (RTOSs) can be used. However, RTOSs can create overhead in systems. Using hardware accelerators can significantly reduce overhead. In this work, we survey the major overhead in embedded systems and identify and analyze some of them in detail. We then propose and discuss nanoprocessors, as configurable hardware accelerators, to lower this system overhead. Our simulation result shows that nanoprocessors can improve system performance at a nominal cost. NANOPROCESSORS: CONFIGURABLE HARDWARE ACCELERATORS FOR EMBEDDED SYSTEMS