Jim Mitchell

Sun Fellow & VP, HPCS Research Program, Sun Microsystems Labs

Dr. Mitchell is a Sun Fellow and Vice President of Sun's High Productivity Computing Systems Research project under contract with DARPA. Prior to this, he was Vice President in charge of Sun Microsystems Laboratories. Before that he was Chief Technology Officer, Java Consumer & Embedded products, which followed his time as VP of Technology & Architecture in the JavaSoft Division. Prior to his involvement with Java Technology, Dr. Mitchell was in charge of the Spring distributed, object-oriented operating system research in Sun Laboratories and SunSoft.

Before joining Sun in 1988, Dr. Mitchell was head of research and development for Acorn Computers (U.K.), where the ARM RISC chip was designed, and President of the Acorn Research Center in Palo Alto, California. In 1980-81 he was Senior Visiting Fellow at the Cambridge University Computing Laboratory. From 1971-84 he was at the Xerox Palo Alto Research Center and was a Xerox Fellow.

Dr. Mitchell has been working with computers since 1962 at the University of Waterloo where he and three other undergraduates developed the first WATFOR compiler. He has a Ph.D. from Carnegie-Mellon University and has worked on programming language design & implementation (Mesa, Euclid, C++, Java), interactive programming systems, dynamic interpretation & compilation, document preparation systems, user interface design, distributed transactional file systems, and distributed, object-oriented operating systems. He has also worked on the design of hardware for computer graphics, high-level language execution, and audio input output.

Towards a Peta-Scale Supercomputer

In July, 2003, DARPA (the U.S. Defense Advanced Research Projects Agency)awarded Sun Microsystems one of three 50M$US contracts for a 3-year research program to develop in an integrated way the technologies that Sun would use to develop a peta-scale supercomputer in the 2010 time frame. DARPA calls this the High Productivity Computing System (HPCS) program.

Sun's notional system that we are exploring is called Hero (because it can be of heroic size). Perhaps its single most important technology is Proximity Communication, which enables chips in a system to communicate without wires by being placed every close to one
another. The biggest impact of proximity communication is that it enables Sun to develop a peta-scale machine with a globally addressable memory, which not only has high performance, but which also increases programmers' productivity by enabling a much simpler
programming model than current, "cluster-like" supercomputers.

In this talk we will describe some of the fundamental technologies in
Hero and how it can increase productivity in technical and scientific computing.