Cornell Loses Supercomputer Bid

CORNELL

In an unexpected turn of events, the Executive Committee of the National Science Foundation (NSF) voted Friday against continuing to fund the Cornell Theory Center as a national supercomputing facility.

Cornell was part of a national competition for funding with three other supercomputing centers: the Pittsburgh Supercomputing Center, The University of Illinois at Urbana-Champaign and the San Diego Supercomputing Center.

The latter two institutions were awarded the NSF funding.

According to Linda J. Callahan, director of external relations for the Theory Center, the final decision was not anticipated because the National Science Board (NSB), which was supposed to vote Friday, did not have a quorum.

Theory Center administrators expected the matter to be settled when the board reconvened in May. But the Executive Committee of the NSF decided to take control of the vote from the NSB. The Office of Government Ethics permitted two members of the NSB to partake in the vote, replacing members of the committee who had been deemed as having a conflict of interest.

Correct Procedure

According to John E. Hopcroft, dean of the college of Engineering and a member of the board, this was the correct procedure for conducting the vote.

"It's unusual, but still the proper way to do business," he said. "The Executive Committee [of the NSF] is empowered to act on behalf of the board."

While the NSF provided almost half of the center's financial support, Norman R. Scott, vice president for research and advanced studies, said the center and the NSF will take measures to make the loss of funding as gradual as possible.

According to Scott, the NSF will provide transitional support for the center for up to two years.

"The result coming as quickly as it did was a surprise," he said. "Still, we at least had an idea it might happen."

The center is also funded by the State of New York, the university and corporate users.

Possible Staff Cutbacks

Scott noted, however, that one of the ramifications of the loss of funding could be staff cutbacks.

"We'll need to look at how to reinvent the center," he said. "That is likely to result in some rearrangements of the Theory Center."

Because the vote came so unexpectedly, administrators are still unsure of the future of the supercomputing center, whether the university will appeal the NSF's decision and what factors were responsible for Cornell's proposal being denied.

"Cornell has several options now," Hopcroft said. "But given that the funding has been terminated, it may be best for the university to stop being a supercomputing center."

Hopcroft outlined several courses of action that the university might take to maintain the center, including trying to get funding from other federal agencies, New York state or major corporations.

But he noted that since the grant recently awarded to the two other centers is for a period of 10 years, it will be increasingly difficult to upgrade the supercomputer and thus remain competitive.

"The maintenance bill on the computer is $1.5 million each year alone," he said.

Appealing the Decision?

Callahan said the university does not know whether it will appeal the decision until the NSF makes the review of Cornell's proposal public.

According to H. David Lambert, vice president for Information Technologies, there was a "difference in perspective between Cornell and the NSF on what should have been stressed" in the request for funding.

"The two centers that received support were the ones that focused on science and not providing infrastructure and support for scientists," he said.

Lambert added that going into the competition, the university thought that this emphasis on education and research would be advantageous.

According to Stephen P. Johnson, executive director for Government Affairs, there was an original NSF competition for super-computing centers 10 years ago, during which the existing four national centers were established, and automatically funded for 10 years.

This past year, the competition was reinstituted to ensure the centers were still providing high performance computing, he said.

Unfair Criteria?

Cornell wrote a request for funding, which proposed a wide program of infrastructure and research, Lambert said. However, during the review the university discussed with the NSF that perhaps the criteria being used to judge their proposal was unfair.

While the NSF noted this challenge, it did not affect the final decision.

"There were a lot of questions about the review criteria being fair," Callahan said.

Scott, while admitting that the removal of funds was a significant loss for the university said he remains optimistic about the future of the center.

"We are looking at it as a transition from being an NSF funded center, to a national center that still does high-end computing, but with different sources of funding," he said.