Published in June / July 2005 issue of Chip Design Magazine
If any one term could be used to describe the current state of the EDA industry, that word might be â€śconsolidation.â€? Everywhere you look, the industry is in the midst of realignment. Analog and digital systems are being combined on one piece of silicon to form system-on-a-chip (SoC) technology. Major tool vendors are consolidating their presence through the acquisition-via dollars or the courts-of other smaller companies. Design teams are consolidating their expertise by reorganizing their development approach (read â€śoff-sourcingâ€?). Following this trend, the IEEE is strengthening its presence in the chip-development world through the formation of an EDA council. Several of the more prominent CAD- and CAE-oriented societies will come together under the umbrella of the Council for Electronic Design Automation (C-EDA). The stated goals of the council will be to â€śhelp coordinate EDA activities, foster interdisciplinary research, offer a way to enable new initiatives, and pursue the recruitment of young talent within the field.â€?
The five charter IEEE societies that make up this council will mirror the functions that are represented on todayâ€™s most complex SoC designs:
At first glance, this council may appear as a hardware-only club. Not so, says Al Dunlop, a longtime IEEE member and EDA proponent who will head the new council. I recently talked with Al about the importance of the C-EDA. He explained that embedded-software developers will play an active role in the council through the Computer Society. Al pointed out that coordinating the activities of all of these different societies is a necessity for the successful design of future circuits.
What changes will the EDA council bring to the chip-design community? Al was understandably hesitant to elaborate on any changes until the council is formally approved in June. But he did note that the council would have influence over the CAD-related conferences that are currently sponsored by the five member societies. For example, DAC-which is sponsored in part by the IEEE Circuits and Systems Society-would be all or partially under the umbrella of the new council. The councilâ€™s influence also would extend to some IEEE CAD publications and perhaps part of the Design and Test magazine (published by the Computer Society).
Another area that might be affected by the council is the development of technical standards. Though no immediate activity in this area is planned, Al agreed that there are quite a few opportunities for participation by the council. But he was quick to add that the C-EDA does not intend to compete with existing standards bodies like the VSIA, Si2, OCP-IP, and others. â€śThese are good groups with the right topics. We want to encourage their work,â€? he said.
All of this activity sounds good. It makes sense to bring these different societies together under one banner. This effort certainly mirrors todayâ€™s design reality, where analog-digital, hardware-software, and local-global teams must play nicely together to meet the demands of time-to-market, increasing design-test complexity, decreasing product cost, and chip size. As those of us with engineering experience well know, however, bringing together such a diverse group is fraught with peril. Add to these challenges the political rivalries that sometimes exist among technical societies, and the C-EDA council will have its hands full.
But there is hope. As head of the council, Al recognizes that each society has its own limited understanding of the bigger picture (read â€śsystemsâ€?). All of them are struggling with the challenge of consolidating everything on a higher level (read â€śhigher level of abstractionâ€?). This problem is not new to Al, who was Chairman of the first Hardware-Software co-design workshop offered by the Circuits and Systems society almost 10 years ago. If anyone understands the problems with multi-discipline and technology projects, itâ€™s him.
But even the best practices of a co-design methodology are only part of a total system-engineering approach. Itâ€™s a good start, but will it be sufficient? For the sake of the EDA community in the U.S., I hope the answer is â€śyes.â€?