@goafrit, no dobut Chip design is not a one man game but its always good to have a leader who can choose the best team. I am sure AMD has enough talented resource pool which along with Keller will make good team.
I'm sure AMD is already taking low power plus high performance chips very seriously, not only for mobile but everywhere. The recent news about widespread power outages in India just helps to drive home the point. Data centers suck huge amounts of power and generate lots of waste heat, so every element of those is getting intense scrutiny regarding power/performance metrics.
Bringing Mr. Keller back into the fold, AMD is sending a clear message that the top management is committed to the work. As for getting the work done, I'm sure there are many very talented AMD engineers already working on high performance plus low power processors. And AMD will attract some more now that Mr. Keller is on board.
Jim Keller is an industry veteran and has been around many succesful designs -Dec Alpha, SiByte, PA-Semi, Apple but he has not been the sole resonsible architect for any of these (except maybe PA-Semi). Crediting him with the low power CPUs and chips from apple is stretching it.
I don't see this making any difference to AMD's short or long term viability.
Keller was one of the engineers behind Opteron, the generation that really made an innovative leap over Intel by integrating the memory controller and a coherent bus (HyperTransport) on a server chip.
AMD needs some of that leap ahead product innovation today. It now has all the pieces in place with leading graphics cores, multiple x86 cores, an ARM partnership--and now a returning veteran microprocessor designer.
Untrue! He had *nothing* to do with Opteron. He has been around succesful designs and is certainly knowledgeable on them but he has not been the chief architect for any of the shipping Apple or AMD parts.
Blog Doing Math in FPGAs Tom Burke 21 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...