One trend occurring over the past decade is a reduction in the overall innovation by entrepreneurs in the semiconductor space. With the roll out of multi-core processors it is obvious that chokepoints in processing speed are going to lie in places like the bus and memory. That is where MetaRam comes into play. A small team like the one at MetaRam, developing next generation memory, promising dramatic performance increases (8GB Dimms and 8GB DDR2 RDIM are just appetizers), is just the type of venture the Valley needs more of. Aside from impressive technology there is a vetted strategic roll-out in place, orchestrated by Fred Weber, former CTO of AMD.
SM: Fred, I would like to start by setting and gaining context in regards to your personal background. Where do you come from and how did you get into all of this?
FW: I have been interested in computers forever. I started programming on cards and paper tape on old PD computers. When I started my school I initially studied physics but ended up changing to computer science because it is a lot easier than physics. I was really good at mechanics but electromagnetism threw me, and computers are much more like mechanics than they are to electromagnetism theory.
I started my career with Gordon Bell doing large scale multi-processor supercomputers. I started at Encore and then moved on to Kendall Square Research which was one of the massive parallel computing machine developers.
SM: What timeframes are we talking about here?
FW: I joined Encore in 1985 and Kendall Square Research in 1987. It turned out that was a tough time to make supercomputers. If you look at 1990 you saw the end of the Cold War and even Cray was having a hard time making money. The reason I got involved with Encore was two-fold. First, a professor of mine told me anything Gordon Bell was doing would be great to get involved with, which turned out to be true. Second, they were doing multi-processor design and the word in 1985 was that Moore’s Law was about to end. Everyone expected that what was going to save us was multi-processors.
My later work at AMD was all around bringing multi-processors to mainstream machines. What you are seeing now with multi-core rollouts is that, finally, multi-processors are everywhere and are the savior. We were just looking at it 15 years too early. A lot of the groundwork we laid then is very similar to what goes on now with the multi-processors. The bottom line for me is that I was really interested in building big, complicated, fast computers using multi-processors.
SM: Where you focused on the hardware or the software sides of those projects?
FW: I started out a software guy writing real-time software. A lot of the hardware guys told me if I liked designing real-time software then I would like designing hardware. It turned out to be true. That is where things started out. I then moved out here to the Valley and started working NextGen with x86 processors. I got courted by AMD and spent 10 years helping AMD get serious in the processor business.