US-Egypt Cooperative Research: Design and Implementation of Automatic Parallel Detection Layer

University Of California-Irvine, Irvine CA

Investigators

Abstract

0004438 Bagherzadeh Description: This award is for support of a cooperative project by Professor Nader Bagherzadeh, Department of Electrical and Computer Engineering, the University of California, Irvine, California and Dr. Hesham Eldeeb, Computer and Systems Department, Electronics Research Institute, Cairo, Egypt. In applications requiring high computing power, one approach is to put together many processors in the same computer. The utilization of all the available computational power involves a tremendous programming effort, which creates a need for compiler and run-time support. Software developers face the challenge of how to exploit application parallelism without rewriting and redesigning current sequential programs and algorithms from scratch. One approach is to isolate programmers from parallelism by adopting optimizing compilers for parallelism detection, named as implicit parallelism. This frees programmers from the details of controlling concurrency, thus concentrate on the application. The work of the two scientists is aimed at fully automating the parallelization process of computer program applications to transform the sequential source code into parallel source code. They will introduce an Automatic Parallel Detection Layer (APDL) which accepts as input a sequential source code C program and produces as output an equivalent source case program in parallel form which is then executed on a distributed memory multicomputer machine. Scope: This collaborative project brings Dr. Eldeeb's experience in high performance computing, parallel virtual machines, parallel logic programming and virtual reality to work in the PI's laboratory at UC, Irvine. Dr. Bagherzadeh's research interest is in the area of high performance computing, superscalar design, multithreading and virtual reality. His expertise also includes star-connected cycles interconnection networks and real-time computer processing. His theoretical and computational background will significantly help in the successful completion of this research. He will be responsible for providing the expertise to formulate the structure of the automatic layer, and will provide the applied cases for the project according to certain required applications in his research laboratory. The project meets INT criteria for support of cooperative projects that are mutually beneficial. Funding for this project is provided by the Division of International Programs and the Division of Computer-Communications Research.