San Diego, Calif. -- Despite what headlines may be screaming, it's the scientific sector, not commercial applications that is the driving force behind high performance computing. Such was the message delivered by Rick Stevens, director of Argonne National Laboratory's math and computer science division and the principal architect of the I-Way, during an interview with HPCwire earlier this week. The I-Way is the system linking supercomputer centers across the nation. (See LIVEwire article 7557 "I-Way; The Next Generation of Internet Debuts at SC'95" 12.05.95). IT'S NOT CASH THAT MATTERS. IT'S IDEAS "You have to look at where the ideas are happening. The few large-scale vendors have very bright people, but they no longer have the research mission -- to do basic research to push the envelope," Stevens said. "What we used to think of as the national brain-trust that was in the commercial R&D sector is being systematically dismantled. And it's being dismantled for economic reasons." As corporations dig deeper into their pockets to compete in the supercomputing arena, and research institutions face uncertain funding, the source of technological innovation comes into question. "The real issue is: Where will innovation continue in the future? The research at scale that we think needs to be done to invent the computer technology and environments for the next century has to be done somewhere. And I don't believe it's going to be done in the labs of the commercial companies who are preoccupied with the short-term," he continued. Although he concedes that the research could be done by commercial companies, Stevens insists that it has to be done by people interested in pushing the envelope -- not just making some incremental step. "The I-Way is about pushing the envelope: pushing it in networking, pushing it in applications, pushing it in software environments, pushing how many computer cycles we can bind together," Stevens said. Pushing the envelope is often best accomplished in a collaborative, vendor neutral environment, and such was the case with the I-Way. "This is a grass-roots activity," Stevens said. "This is not an activity that was heavily funded from the outside. We did get serious support from DOE, NSF and ARPA. We're probably looking at a ratio of real dollars to in-kind dollars of 1 to 8. So in-kind corporate support was necessary to make this work. "The use of computing in science that has given the scientific community the freedom to try new things has always driven computing. And nothing has changed. Where did the Web and Mosaic happen? The Web was invented at CERN and Mosaic was invented at a university. Look at the relative impact of something like the Web and Mosaic versus, let's say, something like Lotus Notes. One can argue that Lotus Notes from a commercial standpoint was improving people's productivity in companies much earlier. And as a commercial product, it's a fine product. But if you want to change the way the planet does its business, we have to adopt a different model. "That model is to demonstrate, to do something, to make that technology available and have things like Netscape happen," Stevens concluded. "And when companies recognize some phenomenon is happening, they'll invest in it. We have to avoid the temptation of moving away from that point."
ARCHITECT OF I-WAY SAYS SCIENTIFIC SECTOR STILL DRIVES HPC
December 6, 1995