Dr. Randal Bryant, Dean of the School of Computer Science at Carnegie Mellon University, has been awarded the 2009 Phil Kaufman Award for Distinguished Contributions to EDA.
I had a chance to speak at length with Professor Bryant, an articulate and thoughtful individual, on September 8th, the day his award was announced. The conversation was both enlightening and stimulating. My congratulations to Dr. Bryant on his award, and to EDAC and CEDA together for their wisdom in choosing Dr. Bryant to receive this commendation.
A Conversation with Dr. Randal Bryant
Issues of Academia …
Q – Let me start by asking you why Carnegie Mellon has been such a profoundly important contributor to EDA?
Dr. Bryant – We have been strong in EDA because of a few important contributors. I started here at CMU 25 years ago, and specifically came here because there were well-known people in EDA on the faculty.
They had formed a center [for research into EDA], which continues to receive a lot of funding from the Semiconductor Research Corporation [SRC], which in turn gets its funding from the semiconductor companies.
Q – Is CMU in competition with the Bay Area, specifically Berkeley and Stanford, for grad students and funding in EDA?
Dr. Bryant – For quite a while, there were two SRC Centers of Excellence in EDA. One was at U.C. Berkeley and one was here at CMU, so we’ve been peers with Berkeley for a number of years. [Naturally] there has always been a bit of rivalry between us, but it’s always been a respectful rivalry.
In addition, I am well acquainted with previous Kaufman Award winners from Berkeley, including Richard Newton, Alberto Sangiovanni-Vincentelli, and Bob Brayton – all of whom I have known well. I have a great deal of respect for Berkeley.
Q – What do you tell young grad students who are in the process of choosing between Berkeley and/or Stanford, and CMU?
Dr. Bryant – It’s always been a little bit easier for the California schools to recruit students, just because they’re in California. Nonetheless, we’ve always been able to attract great students here to CMU. I always think of U.C. Berkeley and CMU as having strengths in slightly different areas, so that is [an additional issue] that potential grad students take into account in choosing a university.
Q – What specific strengths does CMU offer?
Dr. Bryant – The advantage CMU offers is that we have a range of faculty working in EDA, everything from statistical process modeling, to high-level logic synthesis and verification. We offer a full spectrum of coverage [in the different areas in EDA].
If EDA is something that a student is interested in, there are lots of ideas and insights available here [for them to take advantage of]. We are not just narrowly focused in one area. In addition, there’s a nice sort of camaraderie here between our faculty and our students.
Q – Many people believe the EDA ecosystem that fosters startups is centered in the Bay Area. Is that an additional challenge in recruiting students to come to CMU in Pittsburgh?
Dr. Bryant – Yes, it’s true that Pittsburgh is not Silicon Valley, and is not a place that has generated companies and industry.
We have had a couple of EDA-specific companies start in this area, although I think just about all of them have either been acquired or have moved out to Silicon Valley. That’s not a reflection of the technical expertise here, however, but a reflection instead of the [nature of] the Pittsburgh area.
Q – What do you think it is about Silicon Valley that makes it such a draw?
Dr. Bryant – Silicon Valley has generated so much [innovation], and is such a magnet, that to establish a new center that’s not in Silicon Valley is almost impossible at this point.
However, if you take Austin as an example. There’s lots of electronics industry there. Austin offers a favorable climate for business, a well-trained workforce, and a university nearby with strong competence in the area. It takes this whole partnership between industry and university talent working together to [create this type of environment].
Here in Pittsburgh, we have started building a better technology climate and [an environment] where startups can work, particularly in the areas of robotics and the Internet.
Q – Speaking of robotics, and expanding a bit beyond EDA, how does CMU compete with MIT in that area?
Dr. Bryant – It’s true that they have robotics at MIT, but in that area we definitely win. Our Robotics Institute ( http://www.ri.cmu.edu/) is the largest academic institute in the world in robotics. CMU is the only university in the world where you can earn a Ph.D. that specifically says “Robotics” on the diploma. Students come here knowing that.
Again, there is certainly excellent research into robotics going on at MIT, but those studies are spread across electrical engineering, computer science, and mechanical engineer. Here at CMU, we have a complete Ph.D. program that [incorporates] all of that.
Q – Now that you‘ve successfully met the challenge of differentiating CMU from other world-class institutions, let’s turn back to EDA. What do you believe is the future of industry?
Dr. Bryant – In my opinion, EDA is moving up the food chain, away from hardware design and more into implementation on FPGAs, and even the software. That’s been the hardest divide to cross, that mixture of hardware and software together in an embedded system.
Most embedded systems are being designed today with traditional software development tools, work that starts at a high level and is then mapped down to the software and hardware. That [strategy], however has not yet been cracked open with industrial strength tools. There’s a great deal of progress needed here, progress that represents the future of EDA.
Q – Why has there been such a delay in developing ‘industrial strength‘ tools for hardware-software codesign?
Dr. Bryant – Because it’s hard! You have two different modes of people [that you’re trying to bring together], people coming from opposite ends of the problem.
Most embedded systems designers start by deciding which processor to buy, and then they start wiring together the software. Instead, they should be starting with, ‘What are we trying to accomplish with this system?’
To get people to change their way of thinking is always the greatest challenge.
Q – Maybe CMU needs a new Ph.D. program specifically geared towards hardware-software codesign?
Dr. Bryant – Actually, that is a very active area of research at CMU, and at other universities of course.
Q – In a recent interview, Berkeley’s Dave Patterson told me the solution to the parallel-programming problem is not going to come from some 13-year-old tinkering with his gaming system in the garage. The solution’s only going to come from the hard work of many people coordinating their efforts. How about hardware-software codesign? Can we expect a wunderkind to find a breakthrough solution in this area?
Dr. Bryant – I agree with Dave Patterson. It won’t be a 13-year-old grad student, plus we don’t have too many of those. [More importantly], I believe we need to let 13-year-olds be kids for as long as possible.
You can find the full EDACafe event calendar here.
To read more news, click here.
-- Peggy Aycinena, EDACafe.com Contributing Editor.