Issue link: https://iconnect007.uberflip.com/i/1285883
SEPTEMBER 2020 I DESIGN007 MAGAZINE 17 charts. They decided that things would run hot- ter inside the board than outside, and they took half the current. Holden: It's crazy. I went through our design manual and relayed some of the numbers back to you, where you gave me the condition. I said, "The data in this chart comes from HP labs, but there's a reference on the bottom from an Air Force ManTech program. Somehow, HP labs had also gone back to the military." But I couldn't help you because they didn't know anything about the IPC standard. We weren't allowed to use IPC standards. People were running into a fistfight with me because talking about those curves not being correct was totally unacceptable as if they were handed down by God on golden tablets. Nobody knew where they came from, but nonetheless, don't violate them. And it went from one or two pages up to 70–80 in the final IPC-2152. Jouppi: And that's just touching the surface. I had a long list of all the things that I thought that we should still do. But when you're work- ing with volunteers, it goes extremely slow, and only a few find time to contribute. Everybody was overworked during those years. I teamed with the University of Colorado in Denver, and we wrote a grant proposal to the National Sci- ence Foundation (NSF). After being turned down for the NSF grant to fund the work, I end- ed up creating my own lab, collecting data to validate some previous testing, and developing a software tool to manage the data. Some of the data that was collected went into IPC-2152. Holden: I was at Mentor when they purchased the software tool FloTHERM, so I used it be- cause I was working for the vice president, and everyone were either electrical engineers or software people. The software came from the Soviet Union. They didn't write it, but they adapted it from a Russian worker. They asked, "What can we do with this three-dimensional computational fluid dynamics?" I was working with it on a Dell computer, and the finite element analysis would give me real-time animation. I said, "Is this software for real, or is somebody pulling my leg?" because you can't do three-dimensional fluid dynamics on a Dell computer using an Intel processor. It takes a Cray supercomputer to do that kind of work. And they said, "No, the Russians figured this all out because they didn't have Cray com- puters." I responded, "This is really something. I've never seen real-time, three-dimensional fi- nite element computation like this." Jouppi: They did a great job. Russian mathema- ticians are always incredible. Holden: What are you doing now? Jouppi: I am semi-retired. I haven't been tak- ing on any work, and I've been doing a lot of artwork. Out of high school, I wanted to be an artist, but I looked at all these great artists who were starving, and I didn't want to have that kind of lifestyle. I chose engineering to provide a lifestyle that would be reasonable, and it has taken until now to really start digging into my artistic side. Shaughnessy: This has been really good, Mike. We appreciate you taking the time to talk with us. Thank you. Jouppi: It was fun. Thank you. DESIGN007 One of Mike Jouppi's recent pieces of art: a tree stump brought to life as an owl spreading its wings.