At his new job, Jim gets the title of CVP and Chief Architect for CPU Cores, basically what it sounds like. He reports to Mark Papermaster, also ex-Apple, but not ex-ex-AMD or ex-ex-ex-DEC. Then again, Keller isn’t ex-ex-ex-Cisco, Papermaster is.
Jim’s last job was Director, Platform Architecture at Apple, basically leading the A4, A5, and likely others that may be named A with a number after them. He came to Apple from PA Semi when the former bought the latter, and sadly did away with their innovative low power PowerPC line. You can see how large chunks of that thought process found its way into the A-Series of chips, and those are rumored to have done pretty well in the market despite their odd cases.
Before PA Semi, Jim Keller was at SiByte, and parts of that still live on at Broadcom. If you know the MIPS world or network processors, you know why what they did was important, but to the average consumer, it probably didn’t mean much.
A bit before that, Jim co-authored the Hypertransport specification, an I/O paradigm that was way ahead of it’s time. It still holds up reasonably well a decade later, eons in computer design time frames. He also was lead architect on the K8 CPU, and worked on the K7 before that. You might remember those two as being ahead of their time too, far ahead.
Going back to the stone age at DEC, he also got his fingers dirty on the Alpha 21164 and 21264. Those chips were also regarded as milestones in the industry, and given the ties between the later DEC Alphas and the early K8s, well, you probably get how some of that happened now, right?
All in all, this is a good thing for AMD, and not nearly so good for Apple. The litany of execs leaving AMD of late has caused SemiAccurate to say that nothing good can come of this far too often for our liking lately. This time, all we can say is that a lot of good can and will come of this. Jim Keller is one of the good ones, and you don’t leave directorships at Apple without a damn good reason.S|A