An Undergraduate Parallel/Distributed Computing Laboratory

点击翻译按钮获取中文摘要

Through this project, the Department of Electrical Engineering (EE) continues to update and expand its Undergraduate Parallel/Distributed Computing Laboratory. The project supports existing curriculum offerings and widens the use of the technology by using it as a base in other, non-parallel-focused courses, such as Databases, and Artificial Intelligence. Modern computers connected by a LAN can provide a parallel virtual machine with sufficient reliability that the parallel/distributed techniques can become an ordinary, everyday tool rather than just the focus of special interest courses. Old transputer boards have become obsolete and almost impossible to use after 6 years of heavy utilization, and this project is designed to retain the parallel/message-passing paradigm in the EE's curriculum. Many algorithms have portions that could be executed at the same time if facilities to do this exist. Except for expensive special purpose machines, computers have traditionally been SISD (Single Instruction Single Data) processors, executing instructions without the possibility of simultaneous execution. On a parallel machine or facility, such as this new heterogeneous PVM (Parallel Virtual Machine), students are able to execute parallel simultaneous instructions over separate vectors or partitions of data. A modern EE curriculum should include the application of parallel/distributed techniques to problems at the frontiers of computing speed such as image or speech analysis, neural nets, modeling, and graphics. The machines must also be the object of study and be available for student experimentation. A reconfigurable heterogeneous PVM lends maximum flexibility to the laboratory, at modest cost, and uses existing facilities. The refurbished laboratory is immediately available owing to prior expertise in this area, the readiness of space, current computers, and extant LAN facilities. *