Keynote

Lessons from 40 years for Reconfigurable Computing

John Wawrzynek

University of California, Berkeley

Abstract: Since the introduction of the first FPGAs in the mid 1980’s, reconfigurable devices have offered a promising alternative to conventional computing devices. It has long been understood that reconfigurable devices can provide significant advantages over conventional processors; and because of their fine-grain parallelism and flexibility, they can be adapted over a wide range of applications and scale over a variety of problem sizes. However, besides a few notable examples, reconfigurable computing has not reached the wide-scale acceptance that many have expected. The primary challenge has been in ease of use. Past success has only come after custom development of reconfigurable computing platforms followed by laborious hand-mapping and tuning of applications.

Over the years, a major focus of the academic and industrial research community has been on techniques for improving the ease of use of reconfigurable devices for a variety of computing tasks. Many techniques have been tried, including novel fabrics and execution models, standardized HW/SW platforms, new programming paradigms, innovative mapping tools, and algorithm/implementation co-design. While significant progress has been made, more work needs to be done to commercially compete with multi-core processors and GPUs.

Because of their fine-grained flexibility, reconfigurable devices are sure to find an important place in the future ecosystem of computing platforms. However, a question remains. Are reconfigurable devices destined to be everyone’s second best solution, or will they finally find their killer apps?

Speaker: John Wawrzynek is a professor of Electrical Engineering and Computer Science at UC Berkeley and co-director of the Berkeley Wireless Research Center. His research experience is in wireless system prototyping, reconfigurable computing, and design methodology. He has co-founded several successful startups including: Andes Networks—specializing in hardware acceleration of security processing for data centers, acquired by Sun Microsystems, Bina Technologies—building high-performance DNA sequencing
platforms, acquired by Roche Molecular, BEEcube–suppliers of FPGA-based 5G+ wireless prototyping platforms, now a National Instruments company, and currently, Lutris Wireless. He received a BS degree in EE from SUNY Buffalo, a MSEE degree from University of Illinois Urbana/Champaign, and a PhD in Computer Science from Caltech. He was past program chair and general chair of ISFPGA, and has been on the faculty at UC Berkeley since 1989