What Would Joe Do?
Peggy Aycinena is a freelance journalist and Editor of EDA Confidential at www.aycinena.com. She can be reached at peggy at aycinena dot com.
Remembering Steve Levitan: Scholar and Gentleman
April 14th, 2016 by Peggy Aycinena
I had a chance to interview Dr. Levitan in late 2006, as he was ramping up to serve as General Chair of the 44th DAC, and found him to be very sincere and down-to-earth. He was clearly one of those rare individuals who respected the balance between academia and industry, and how each sphere plays an equally critical role in pushing the envelope in electronic design automation. The text of that interview is available below.
Earlier this week, I received a note from Soha Hassoun, Professor and Chair of Computer Science at Tufts and General Chair of the 51st DAC in 2013. Professor Hassoun said that she and Penn State CSE Professor Mary Jane Irwin have written a very nice article memorializing Steve Levitan, set to be published in the May/June issue of IEEE Design and Test magazine.
Dr. Hassoun also noted that she was among many colleagues who were deeply moved at the sudden news of Dr. Levitan’s passing: “I had been working close with Steve, Jinjun Xiong and Eli Bozorgzadeh during the past few months to organize the Young Faculty Workshop at the year’s DAC. The news of his death saddened us all. Steve was an outstanding mentor, a great role model and incredibly supportive of volunteer initiatives within the EDA community.
“In 1997, while a graduate student at the University of Washington in Seattle, I had the idea of creating a safe space for Ph.D. students to interact with academic and industry researchers. I proposed a poster session format, and Steve invited me and Olivier Coudert, who volunteered to co-organize this event, to attend the Fall ACM/SIGDA meeting to pitch the idea to the SIGDA board. I recall how Steve made us feel welcome at the meeting and valued our ideas.
“SIGDA decided to support the idea, and that was the beginning of the what is now known as the Ph.D. Forum at DAC. The Forum is now in its 18th year, with its own technical program committee and over 100 annual submissions with hundreds of attendees. I will always be thankful to Steve for opening that door for me, and allowing me and many others the chance to contribute and engage with the EDA community.
“Steve’s presence will be missed at every DAC and ICCAD event for the foreseeable future, but his spirit will stay with us for a long time to come.”
The following excerpt is from the obituary for Steven Levitan, published in The University Times.
“An award-winning teacher and researcher, Steve Levitan was the John A. Jurenko Professor of Computer Engineering in the Swanson school’s Department of Electrical and Computer Engineering (ECE) and held a joint appointment in the Department of Computer Science. He worked in the design, modeling, simulation and verification of highly parallel systems, focusing on parallel and optical computer architectures, VLSI systems and mixed-technology microsystems.
“His research interests centered on integrated circuit technology, including designing and modeling nontraditional circuit technologies and neuromorphic computing. Among other research, Levitan was co-PI on a National Science Foundation-funded visual-cortex-on-silicon machine vision project.”
[First published 7 December 2006 in the DAC Newsletter]
DAC 2007 General Chair Steve Levitan is no stranger to the mysteries behind the magic of the Design Automation Conference. He’s been on the DAC Executive Committee for 9 years – 4 years in his capacity as Chair of SIGDA [ACM’s Special Interest Group for Design Automation], 3 years as Multi Media Chair, a year as the New Initiatives Chair, and a year as Vice Chair of the conference. Now entering his tenth year on the Executive Committee, Levitan is General Chair of DAC and there are few mysteries left for him.
He knows the event requires careful coordination of hundreds of volunteers and tens of thousands of hours of effort, a delicate balancing act between the needs of academia and those of industry, and a constant vigilance to guarantee that the latest and greatest in technology are adequately showcased during the week. He is no stranger to the commitment, sacrifice, and tedious attention to detail needed to make the conference happen.
Nonetheless, if you have a chance to speak to Steve Levitan about DAC, you’ll quickly see that he still firmly believes in the magic. He believes when thousands of people come together in a single venue to discuss everything EDA, the whole is greater than the sum of the parts, and the industry and its technology take yet another tangible step forward to the benefit of all involved.
Levitan says, “It’s simply amazing to see the passion that people bring to the work of making the Design Automation Conference a reality. It takes a lot of people and a lot of effort, and it’s an all-volunteer organization. There are 16 members on the EC, while the Technical Committee for DAC 2007 has over 80 members. Adding to that headcount, there’s the Panels Subcommittee, the Tutorials Group and Subcommittees, the Exhibitor Liaison Committee, and the Strategic Planning Committee. And that doesn’t even touch on the number of people from MP Associates, who manage the day to day operations. It’s just a huge undertaking.”
DAC 2006 in San Francisco logged over 10,000 attendees and DAC 2007, scheduled for early June in San Diego, could be even larger. Levitan says, “Both the total number of paper, panels, and tutorial submissions and the number of companies participating in 2007 should be up compared to 2006, which was already at the highest participation level in years.
“Early indicators suggest there are lots of new and interesting ideas to be explored at DAC. Although, of course, what the hottest topics at DAC 2007 will be is not yet settled, but you can certainly see some of them reflected in the various Special Sessions proposals being considered by the EC for the 2007 program.
“We’ll know better in a couple of months, but it wouldn’t be surprising to see topics of evolutionary advances in CAD, new ways of looking at the die, at fabrication technology, and DFM. Also, statistical modeling and robust, variation-tolerant design will undoubtedly be highlighted, as well.”
Although Levitan is an academic, the John A. Jurenko Professor of Computer Engineering in the ECE Department at the University of Pittsburgh, he is fully aware that one of the biggest challenges at DAC is to maintain and support both the commercial and academic aspects of the conference.
He says, “Although the hot topics at DAC are often forward looking, it’s also important to remember that many of the technologies are being commercialized right now. And that process adds to the fascinating dynamic at DAC. It’s an extremely complex conference, because it has such a diverse audience.
“There are a lot of people on the exhibit hall floor involved in industry who may not make it upstairs to the technical sessions, while there are also a lot of people upstairs from academia who don’t appear to give a whit about what’s going on downstairs. It can seem as if these two communities are disconnected, but in reality the best of each community is fully aware of the best of the other community. And that’s where the magic of DAC resides.
“Every year, several of us who are academics make a point of walking the exhibit hall floor and looking at the new tools from each company. We ask ourselves, where did this work come from? Was this a technology that was developed in-house, did it come from a company that was acquired, or was this so-and-so’s work at a particular university? Those of us who participate in this exercise are really conscious of the whole food chain in EDA and the history of the companies, not to mention the Ph.D. theses driving this or that technology.
“By the same token, people from the companies – Cadence, Synopsys, Mentor, Magma, and so on – are often upstairs sitting in the technical sessions taking copious notes during the presentations. There’s a synergy that takes place at DAC that definitely defines and defends the ongoing need for this conference.”
Levitan says guaranteeing the quality of the technical program is crucial to the synergy and to maintaining the pre-eminence of DAC within the research community: “We are confronted with a challenging situation with regards to the technical portion of the program at the Design Automation Conference. Some academics believe that a paper published in a print journal is worth more than a paper presented at a conference, and that may be true for some conferences. But that’s not true for DAC.
“The review process that goes into selecting the papers presented at DAC is equal to, or surpasses, the peer-review process involved for journal publication. The DAC quality selection process is directly linked to the caliber of our large technical review committee, and the amount of time they take to evaluate each submission.”
Even more crucial than the technical program itself, however, is the research food chain that produces the submissions in the first place. Levitan says that the quality and quantity of that research hinges on the ability of the universities to attract bright, innovative graduate students and the funding to support them.
He notes, “There is an evolutionary nature to this process. People do interesting work in academia, which often becomes the basis for a startup, which then evolves into a full-blown company or an acquisition into an existing EDA company. So there’s a path from the graduate students – funded by the SRC, NSF, DARPA, or the CAD companies themselves – into industry. The grants for these students are reviewed not just by academics, but by industry people as well, which means your work in academia is being evaluated by the industry year-round, not just at DAC.
“Also, many grad students do summer internships at the CAD companies, and often go on to work for those companies as developers after graduation, or as tool users in the customer companies. So from the very beginning, there’s been a tight relationship between academia and industry within the EDA world. It’s quite unique.”
“Unfortunately today,” he adds, “that whole evolutionary process has gotten quite difficult. It’s hard these days to get access from within the universities to the state-of-the-art technologies being developed – things like 90-nanometer and 45-nanometer production processes – either because of NDAs, or because the universities discourage research work that comes with restrictions.
“This is the type of work that a company might fund, but might not want openly published. Many universities have found creative ways to get around these barriers, perhaps by taking their research farther afield than what the mainline companies are looking at right now.”
Nonetheless, Levitan notes that many universities really want to be looking at the current, cutting-edge problems. They don’t want to be working on esoteric technologies with minimal practical applications. He says, “We want to do real work on real-world problems, because that means our students can have access to practical examples and applications for their work. They spend a lot of time solving problems and want to see if the solutions they develop are useful.
“Basically, the real issue is that university professors just want to be successful. They can’t work by themselves. They can’t write 100,000 lines of code by themselves to solve a problem, so they need to have a research group. But to build that group they have to ask themselves, does the money come from Federal grants or from industry?
“There are differing opinions on all of this, but ultimately the ability to create and sustain a research group is important. It’s important for the tenure of the professor, and it’s important for the graduate students who need the funding to attend school in the first place. Ultimately, of course, it’s important to industry that this process continues uninterrupted.”
Levitan says that today many administrators are encouraging an entrepreneurial attitude within the university setting. Senior faculty members are encouraged to teach the junior faculty, who in turn need to teach the graduate students, that the university is not just an ivory tower. Faculty and students alike need to develop and market their ideas to secure the necessary funding to pursue them.
Per Levitan, “The process that professors need to master is one of couching their research processes and their expected results in a way that people outside the university can really understand.”
And that, he insists, means the universities have an obligation to commercialize their technology. Which brings us back to DAC.
Levitan is excited when he says it’s at DAC that the universities are able to efficiently showcase their nascent technologies and emerging research stars in close proximity to the bulk of the commercial players within the industry. It’s at DAC that a company who might license or fund a technology coming out of a university has a great opportunity to hear about projects that have potential applicability to their own portfolio of tools or in-house research initiatives. It’s also at DAC that researchers from industry who have worked side-by-side with researchers from academia enjoy the opportunity to showcase their joint efforts.
All of this, Steve Levitan says, is part of the dynamic, and the dynamism, that fuels a widespread commitment to DAC, across-the- board from industry and academia. Levitan is unequivocal in his belief that the magic of DAC resides in its impact on the evolution of the tools for electronic design automation. He believes that his efforts, and the efforts of the countless others who are working to make the next conference a success, are important and the results will be substantive.
There are no mysteries to the amount of work needing to be done between now and next June to make the Design Automation Conference in San Diego a success. No mysteries, but in the end there will be magic. And starting on June 4th, 2007, DAC General Chair Steve Levitan will be waving the wand.