"What could Celera possibly need a 2-teraFLOP computer for, when they're out of the genomics business? ... With IBM, it's probably just too little, too late. I think Compaq made very good hay out of their cooperation with us on sequencing the genome. I think they went from being behind to ... being sort of the dominant player in high-performance computing in the life sciences, because of that association.
"It was a good association for us ... I didn't know how to evaluate high-end computing. I'm not a computer scientist, I'm an experimental biologist! So I set up an experiment. I had IBM and Compaq bring in their best machines, and we ran an experiment. We gave them a bacterial genome to assemble in our assembler and see who could do it faster. And the [Compaq] Alpha chip was able to do it about three times faster.
"I was personally very disappointed [that Compaq abandoned Alpha]. But again, I'm an IT consumer, not an IT expert. We could not have assembled the human genome without it, and so I developed a certain affection for the Alpha chip ... the Alpha was truly a wonderful technology.
"What we did with Compaq and Celera was a critical, essential experiment for the time that there is very good reason never to repeat again. We built a special-purpose computer for assembling the human genetic code under the conditions that we had. Even with all the 1.5 tera-FLOPS of capacity that we had there, for many of our things we were still compute-limited. But the cost of that facility ... I mean, the room the computers were in was a $6-million room ...
"Each hospital, each clinical center has to have the compute capacity to deal with its patients' population's genetic codes. Not as the be-all and end-all of all medicine, but it's going to contribute to the understanding of every bit of medical care in the future. So we have to have a cheaper, replicable system."
Back to Buying Power