Systems Trends and their Impact on Future Microprocessor Design

Tilak Agerwala
Vice President, Systems
IBM Research


Important application domains like simulation, games, and content distribution are placing new demands on microprocessor functionality and performance. Furthermore, future microprocessors will be designed for a very broad range of systems: highly parallel supercomputers, backend servers, desk top systems, and a variety of embedded systems . With hundreds of millions of transistors available to the microprocessor designer today, it is important to identify the functions that will eventually migrate into the processor chip. A high level of integration leads to a host of issues, including balancing on-chip power consumption with high frequency design, overcoming design complexity, achieving system reliability, and breaking the memory performance barrier.

The talk will describe emerging applications and industry trends for different categories of systems, discuss some of the exciting research and development challenges these systems will impose on future microprocessor design, and give examples of advanced work at IBM in these areas.


Dr. Tilak Agerwala is Vice President, Systems at IBM Research and is responsible for developing the next-generation hardware and software technologies for personal systems, servers, supercomputers, and embedded systems. He is acting-Vice President, Emerging Business, in Research, responsible for bringing research innovations more quickly to the marketplace and for bringing marketplace forces into sharper focus for researchers. Prior to that, he was Vice President, Technical Alliances in Server Group, where he was responsible for improving the eServer product portfolio by leveraging external technologies and innovations, maximizing income through technology and patent licensing, and for moving IBM technologies to market aggressively through the OEM channel.

From 1999 to 2000, Tilak was Vice President of Unix Marketing and Product Management with responsibility for strategy, marketing, market development and product management for IBM's Unix operating systems initiatives. He built important relationships with key industry partners and established AIX as a strategic Unix operating system. He developed and helped implement IBM's game changing Unix/Linux strategy.

Tilak was director of future systems technologies from 1989 to 1992 and was part of the executive team that brought IBM's first RISC workstation to market. From 1992 to 1997 he was responsible for the systems architecture and technology strategy of the RS/6000 SP, the most successful parallel computer of all time. From 1997 to 1998 he was director of server architecture and systems strategy in IBM's server development organization with responsibility for parallel computing, clusters, and systems technologies.

Tilak joined IBM at The T.J. Watson Research Center, where he established and led research programs in artificial intelligence, high performance microprocessors, and parallel computing. Working with John Cocke, he developed the architecture and machine organization of a very high performance, novel microprocessor. This approach became the foundation of the successful RS6000 workstation and established a key industry direction.

Tilak received the W. Wallace McDowell Award from the IEEE in 1998 for "outstanding contributions to the development of high performance computers". He was a member of IBM's Corporate Technical Committee from 1987 to 1989. He is a founding member of the IBM Academy of Technology and served on its Technology Council from 1990 to 1993. He is a Fellow of the Institute of Electrical and Electronics Engineers and a Member of the Association for Computing Machinery.

Tilak received his B.Tech. in Electrical Engineering from the Indian Institute of Technology, Kanpur, India and his Ph.D. In Electrical Engineering from the Johns Hopkins University, Baltimore, Maryland.