ACM is an active sponsor of DAC Gabe Moretti, Editor
The Association for Computing Machinery (ACM), founded in 1947, has over 80,000 members worldwide and has been a sponsor of the DAC from its inception through ACM’s Special Interest Group on Design Automation (SIGDA). During the conference, SIGDA sponsors three events that provide opportunities for students and academics to communicate with EDA professionals. The interactions provide valuable input for both sides of the discussions. They guide students in their choices of research topics and career planning, and they give EDA professional a close look at the most promising research areas pursued by universities worldwide.The SIGDA/DAC University Booth
In its twentieth year, the SIGDA/DAC University Booth is the oldest activity sponsored by the society. The University Booth provides an opportunity for the university community to demonstrate new EDA tools, design projects, and instructional materials at the Design Automation Conference. Attendees of the University Booth can talk to students one-on-one, which may be useful for recruiting purposes. The University Booth also provides space for the presentation of EDA vendor literature and programs of interest to the university community. Student participants in the University Booth are typically provided with travel grants to defray the cost of attending the Design Automation Conference. Last year, the booth had nearly 50 demonstrations from all over the United States, Canada, Brazil, Japan, Korea, France, Singapore, and Taiwan. ACM/SIGDA, EDAC, the Design Automation Conference, and the EDA Industry sponsor the SIGDA/DAC University Booth. Students wishing to participate in the booth must submit proposals by April 2, 2007. In honor of the twentieth anniversary, in 2007 SIGDA will be working with Mentor Graphics to redesign the booth for a new look. This year, posters will be required to be submitted prior to DAC with a deadline of May 1, 2007. Coordinators for this year’s SIGDA/DAC University Booth are Prof. Alex Jones and Prof. Jun Yang of University of Pittsburgh.The Ph.D. Forum at DAC
The ACM/SIGDA Ph.D. Forum at the Design Automation Conference will celebrate its tenth anniversary in 2007. The Forum is a poster session hosted by SIGDA for Ph.D. students who are nearing graduation to present and discuss their dissertation research with people in the EDA community. Unlike technical sessions at DAC, the Forum facilitates one-on-one contact with students. It has become one of the premier DAC events for Ph.D. students in design automation to get feedback on their research and look for industry jobs, while industrial participants get a comprehensive snapshot of academic work in progress. 400-500 people have attended recent forums and this year’s
event should be at least as popular. The Forum takes place on Tuesday evening and includes a complementary dinner provided by SIGDA. It is open to all members of the design automation community and there is no fee to attend. While the forum is co-located with DAC, DAC registration is not required to attend. This year’s Ph.D. Forum at DAC Technical Program Committee Chair is Prof. Tony Givargis from University of California at Irvine. The SIGDA Board liaison for the Ph.D. Forum is Dr. Robert B. Jones from Intel. Design Automation Summer School at DAC
An Insider's View – DAC & The Evolution of CAD
by Peggy Aycinena
Limor Fix has never worked for an EDA company, but she is Vice Chair of the 2007 Design Automation Conference -- and for good reason. She has been intensely involved throughout her career in evolving the technologies at the core of electronic design automation, both in her former capacity as head of an internal CAD R&D group at Intel, and now as director of an Intel research "lablet" in Pittsburgh located at the campus of Carnegie Mellon University
.Intel has three lablets in North America -- one at CMU, one at U.C. Berkeley, and one at the University of Washington. Limor heads up the facility at CMU, and says the group thrives on innovative contributions from top-notch professors and grad students from university.
Limor says, "In each lablet, there is a small group of Intel employees collaborating with the local university in various technology domains, areas that are important to Intel now, or may become important for Intel in the future. This is a very interesting job for me, one that allows me to look at quite new and exciting ideas. The set of people we have in the labs is very strong, and we are constantly recruiting people from the best universities to join us as Intel employees. Therefore, for every project, we have 2 or 3 fulltime Intel employees plus 2 or 3 professors and PhD students doing research on a topic. In addition, each summer we have 10 to 15 interns in the group. Overall, the return is significant because, although each project may have a small number of Intel employees, we accomplish a lot of research through this collaboration."
"Our current topics in Pittsburgh are quite exciting. Among them, we are developing tools and methodologies for parallel software for multi-core products. Many of the new CPUs being introduced -- ones from Intel and AMD, for instance -- have multi-CPUs on the die. Therefore, the way software developers write code has got to change. Because if you're just going to run your old software on a mutli-core PC, you're not going to enjoy the additional power of the newer configuration."
"We are also working on a related project to develop parallel Satisfiability (SAT) solver, an important building block for formal verification, synthesis and other CAD applications. Our new parallel SAT solver runs very efficiently on multi-core machines. From a future-EDA perspective, the entire body of CAD software will need to migrate onto these new multi-core machines because they'll be everywhere. Going forward, for instance, Intel is going to ship multi-core machines, so migrating commercial CAD software in an efficient way is going to be very important for the EDA vendors and the users. The EDA companies are recognizing that very soon everyone will want to migrate their tools, so there's a lot of gain to be had. The work in our lab is indicating that by doing the proper migration and using better algorithms we can achieve significant speedup."
"Microsoft is also making a big effort in this area, looking at parallel compilers. The idea is that the programmer will be able to write sequential code and the compiler will translate it into an optimized parallel executable. It's a problem today that many people are looking at, although as yet it's not clear which will be the right way to eventually solve it. One particular approach is to use transactional memory to give the programmer the illusion of writing sequential code, while the software under the hood makes sure it is executed correctly and efficiently in parallel. These are exciting problems with direct applicability to EDA."
"But these are not the only projects we're working on in Pittsburgh. We're also working in the area of machine vision, analyzing skin or MRI images to assist in the diagnosis of malignant tumors. Our tools search a huge database of images to help confirm a diagnosis. It's a type of search engine -- not Google-like -- but a search engine nonetheless, and just one of several projects in robotics and virtual reality that we're working on. All of our projects in the lablet involve topics that are very hot today, topics that are the new horizons for computing"
As enthused as Limor is about her group's work at Pittsburgh, this is not the first time she's been excited about the technology focus of her research. In the years following graduate school, the first phase of her career was spent with Intel in Israel developing Intel’s formal verification tools.
Limor says, "I earned all of my degrees at Technion in Haifa, a BS, MS, and PhD. My thesis was on formal verification, which at the time was quite theoretical. Model checking was just beginning at the time, and it was only after that technology started to mature that formal verification began transferring from academia into industry."
"Following graduation, I spent two years at Cornell University doing a post-doc. I was then recruited back to Israel by Intel where I lead the formal verification group for 12 years. First we were just a research group, but then our prototype tools became good enough for internal use. The development took a lot of effort, but even more effort was needed to convince the Intel design teams to change their verification methodologies. Integrating the tools into the design flow took several years, but as the designers learned to use the new tools, they began to understand their value and the right way to use them. Eventually, we all learned that we needed to work together closely to discover where formal verification fit into the overall flow."
Limor notes that Intel was not alone during those years in working on formal verification. IBM was studying the technology, as well. She says, "Both Intel and IBM published papers about formal verification at the time -- I'm talking about the 1994 timeframe here -- and then the EDA companies joined the effort. Later the EDA companies themselves started offering such tools, but prior to that it was impossible to buy commercial tools."
Things have changed radically today, according to Limor: "Verification technologies are being widely used for many applications -- some for equivalence verification, some for formal verification of assertions, some for microcode verification, and some for low-power design verification. In general today, third-party tools are now good enough that, in some applications, there is no advantage to using internal tools. In fact, some of the EDA offerings in formal verification are excellent."
She adds, "The certification of a formal specification language was also a huge and important effort required to reach this goal. The process involved a great deal of work from Accellera, working with language donations from IBM, Intel, Verisity, and Motorola. The Accellera committee looked at these offerings for several years and eventually helped make PSL the IEEE standard. That work, of course, helped the EDA companies to align their tools and offerings to the same specification language, which in turn expedited the development of even better EDA formal verification tools."
The current success of EDA verification tool vendors, doesn't mean the internal CAD groups aren't still hard at work, however, according to Limor: "With some of the newer applications, we again see the phenomenon that companies like Intel are using tools that the EDA vendors are not providing. You're still seeing technology leadership from the internal CAD groups, because we continue to have better access to feedback from the in-house design teams who use the tools. We are able to discuss their latest challenges with them and find new ways to apply the technology to their current problems. Again, this work eventually will become common knowledge and then the EDA industry will get involved."
Which is where DAC plays a vital and ongoing role. Limor says that DAC provides a venue for discussion of the cutting-edge technologies coming out of the universities and the user companies, which in turn drive the evolution of EDA tools. As a long-time employee of a user company, now serving in her 5th year on the DAC Executive Committee, Limor says she understands this process well.
Limor was initially invited to serve on the Technical Program Committee for DAC because of her work at Intel on formal verification. Then she was invited to serve on the Executive Committee -- first as Technical Program Co-Chair for 2 years, and then as Exhibit Liaison Chair for 2 years. Even in her 5th year, she continues to admire the work that goes into the conference and the results that come out.
Per Limor, "The fact is that everyone comes to DAC and everyone contributes to creating a special event. We have enormous contributions from academia --- more than 100 people devoting their time to filtering and selecting the technical papers -- and then there are the committees that help to select the tutorials, the panels, and the special sessions. All of the EDA companies, plus their customers big and small, contribute their expertise in this decision-making process. The end result is something that is quite unique and capable of moving the entire design automation domain forward. DAC is really a meeting place where people from industry express their needs, and where they can give talks and participate on panels to fully explain what their current problems are. Academics and EDA professionals benefit greatly from this feedback."
As far as her own responsibilities for DAC 2007 are concerned, Limor says she'll be attending to more specific tasks than just promoting the research agenda of her employer, one of the largest EDA users in the world. She says she'll be focusing on her responsibilities for the strategic development of DAC, but even more importantly, "Ellen Sentovich [DAC 2006 General Chair] and I will be trying to help Steve Levitan [DAC 2007 General Chair] as much as we can. We will try to share the burden with him, just as we know he'll help me next year when I am General Chair at DAC 2008. Ellen, Steve, and I all know that as long as you have strong people around you, even a task as large as putting together DAC is doable."
"But even with our responsibilities at DAC," she insists, "Each of us needs to continue to attend to our day jobs. It's only by working in the trenches on a day-by-day basis, developing or using automated tools for electronic design, that we are able to fully understand what it is people need today and tomorrow by way of CAD tools. People who have not been involved in design automation, or are no longer involved, simply cannot see the community from the inside as we see it. It's difficult to hold down that day job and still do an adequate job in assembling the conference, but we know the DAC machine is crucial for moving the entire industry forward. It's a fantastic show and one that helps to keep the market on track!"
Peggy Aycinena is Editor of EDA Confidential and a Contributing Editor to EDA Weekly.
Details and the submission form are on the DAC web site
Couldn't make all the keynotes and paper sessions? No problem. Selected DAC session videos and 43rd Best Paper winners are now available!