NSF research funds could benefit EDA—or not

At DAC last week Jeannette Wing, assistant director in charge of the National Science Foundation Computer and Information Science and Engineering Foundation (CISE) spoke to a packed luncheon audience on the subject of the NSF and the EDA industry. She outlined the budget her organization spends on research and gave a broad overview of the research areas the organization funds. The good news is that there still is federal money for basic research. The not-so-good news is that it’s not much of a war chest, given the size of the global challenge facing US technology industries, and it’s intentionally not directed at industry development projects.

Wing said that her organization’s role is to encourage research on the basic questions that underlie future developments in industries such as EDA. "We try to find a balance between curiosity-driven research, which could take us anywhere, and research oriented around grand challenges," Wing said.

The approach is noble. But then there’s the matter of money. Wing’s annual budget for grants is well under $600 million. That sounds generous, until you hear a few other facts. Wing said that her budget, tragically, accounted for 86 percent of the total the US government spends on research in computer science. It’s just about the whole show.

In that light, it’s not such a large pot. The CISE research budget is only somewhat larger that the annual R/D budget for one EDA company, Cadence, which expects to spend $450 million this year. To put the number in a broader context, all of the money that the federal government will invest in computer-science research–supporting strategically vital industries from microelectronics to EDA to information services to electronic warfare–is about a third of one percent of the estimated $180 billion we have so far poured into one failed company, insurance debacle AIG. But then perhaps a restructured AIG can keep us on technological parity with the rest of the world.

Research has not been altogether forgotten in the new administration’s thinking, however. Wing said that her budget was augmented last year by $235 million in economic stimulus funds—a lavish three ten-thousandths, more or less, of the $750 billion economic stimulus package. It is not clear yet whether further stimulus dollars will be available for the next year.

These funds get spread over an incredible range of questions, according to Wing. First, the executive—who is also President’s Professor and head of the Computer Science Department at Carnegie Mellon University—explained that on the whole, CISE is divided into three departments: Computing and Communications Foundations, Computer and Network Systems, and Information and Intelligent Systems. Each of these departments administers its own programs under the direction of the assistant director.

Wing then went on to talk about the research questions CISE funded. She said one broad area was what she called Cyber-Enabled Discovery and Innovation. Roughly, this is the application of what Wing called "computational thinking" to advance non-computing areas of science and engineering. Another broad area is in science and technology beyond Moore’s Law: specifically, multicore programming, nano- and bio-computation, and quantum computing. Major initiatives CISE is funding, sometimes with up to $2 million per grant, include research into computing sustainability, intractability in computing problems, molecular-scale computing, and a theory for programming the mobile Internet.

Wing also made it clear that the idea of breaking down the barriers between the silos in the computer science world was dear to her heart. To this end, she described a number of what she called "cross-cutting programs" that are intended to be interdisciplinary. Wing listed three: data-intensive computing, which includes work on data-driven architectures and cloud computing; cyber-physical systems, encompassing everything from smart vehicles to sensor networks, robots, and embedded medical and human-assistive devices; and network science and engineering, which Wing described as an attempt to put a theory around the nature of large networks so that they could be analyzed, modeled, and, ultimately, understood.

Clearly some of these areas are irrelevant to the EDA industry. But some, such as the work in data-intensive computing and network science, could have direct impact on the computing infrastructure upon which large EDA tools depend. And research beyond Moore’s Law could create whole new categories of systems that would require design automation. In an era in which corporate R/D has become for the most part either pure development or simply history, and in which academic research is increasingly seen by universities as a profit center rather than a societal obligation, the existence of an organization devoted to asking questions and inviting proposals on any scale whatsoever has to be good news.

Yet the NSF’s lightly-funded approach to basic science, harking back to the days when science was a rich man’s hobby, not a national priority, stands in stark contrast to the massive funding and formidable organization that the US has in the past directed at strategic objectives such as the creation of the Trident submarines or—less obviously strategic—placing a person on the moon. Perhaps it’s time for us to recognize that technological leadership is not a gentile pastime for the United States, but a survival imperative, and that we must begin to treat it accordingly.