Apple chip designer Jim Keller heads back to AMD

[Jos]

AMD announced today that Jim Keller, the veteran engineer who led development of their successful K8 architecture over a decade ago, will be rejoining the company as corporate vice president and chief architect of AMD’s microprocessor cores. He is leaving his post at Apple where he helped lead development of the A4 and A5 system-on-chip used in iOS devices and the Apple TV.

Keller will be reporting to chief technology officer and senior vice president of technology and engineering Mark Papermaster, a former Apple employee himself who left in 2010 after the whole ‘Antennagate’ debacle. Keller’s efforts will be aligned with AMD’s “ambidextrous strategy” developing both high-performance and low-power processor cores that will be the foundation of AMD’s future products.

"Jim is one of the most widely respected and sought-after innovators in the industry and a very strong addition to our engineering team," said Papermaster. "He has contributed to processing innovations that have delivered tremendous compute advances for millions of people all over the world, and we expect that his innovative spirit, low-power design expertise, creativity and drive for success will help us shape our future and fuel our growth."

Indeed, Keller has an impressive resume. He worked at DEC in the 80s and 90s, where he co-architected two generations of Alpha processors. Between 1998 and 1999 he served at AMD and in that short period he co-authored the x86-64 specification, the Hyper-Transport specification and was part of the system engineering team behind the K7, in addition to his aforementioned role on the K8.

Keller also worked at SiByte and Broadcom as chief architect for a line of scalable, MIPS-based network processors that supported 1Gig networking interfaces, PCI and other control functions. In 2004, he moved on to PA Semiconductor, where he led a team that developed a networking system-on-a-chip with an integrated PowerPC processor, and finally he arrived at Apple as part of the PA Semi acquisition in 2008.

Keller’s hiring follows a couple of high profile departures at AMD, including VP of strategy Patrick Moorhead and Product Director Carrell Killebrew back in December. Just last week, Bob Feldstein, the man responsible for brokering deals between AMD and console makers left to work at rival Nvidia, and prior to that in July chip expert John Bruno who worked on AMD's Trinity APUs departed the company for Apple.

Permalink to story.

https://www.techspot.com/news/49611-apple-chip-designer-jim-keller-heads-back-to-amd.html

 

[ess333]

The question is whether AMD processors will be competitive again after so much pummelling by Intel since Mr. keller has returned. I loved the K8. I bought AMD CPUs for the past 10 years. Recently changed to Intel Sandybridge processor. I really want AMD to succeed so I can go back to buying their products and it is good for the market so Intel won't give us stupid prices. If there is no competition, we know how that turns out: innovation stalls and prices rise

 

[EEatGDL]

I hope to see some pretty good architecture at decent prices from AMD, I've owned almost all my life Intel and I regret having Pentium 4 and Pentium D (both SK775). I sold my PD 950 @3.4 GHz to a neighbor with a lot better SK775 mobo, Hyper-X DDR2 memories in double channel @800MHz, WD Raptor HD 74 GB and a lot of better stuff... and his old PC with a basic SK775 mobo, single channel DDR2 memory @533 MHz, 80 GB @7200 RPM HD and his old Pentium D @2.8 GHz outperformed my PD in the already mentioned better rig in almost all benchmarks we ran (Sandra, SuperPI, PCMark, etc.). Simply stupid, still a question of why that happened I've been wanting Steve to answer all this time.

But anyway, I hope to see a lot better from AMD in the near future, 'cause if they're worth it I'll be going back to AMD [my only PC with AMD was the first I had with an AMD K6].

 

[SNGX1275]

The question is whether AMD processors will be competitive again after so much pummelling by Intel since Mr. keller has returned. I loved the K8. I bought AMD CPUs for the past 10 years. Recently changed to Intel Sandybridge processor. I really want AMD to succeed so I can go back to buying their products and it is good for the market so Intel won't give us stupid prices. If there is no competition, we know how that turns out: innovation stalls and prices rise

Yeah. The K7 was arguably more important than the K8 though. Prior to the K7 AMD couldn't compete with Intel. The K6-2 was sort of a P2 competitor I think (this is about as far back as I know much of anything with 'pc' processors) but the P2s still won. But then the K7 and the Athlons came, and they finally took the price/performance and also performance titles away from the P2/P3/P4 at the time.

Also, disclaimer, before anyone jumps down my throat. I'm not a CPU arch historian, I'm just posting based on what I remember from about 1998-2003 when I was in college as an undergrad and living with a ton of people that were buying computers, sometimes for the first time.

 

[captainawesome]

[quote="
Also, disclaimer, before anyone jumps down my throat. I'm not a CPU arch historian, I'm just posting based on what I remember from about 1998-2003 when I was in college as an undergrad and living with a ton of people that were buying computers, sometimes for the first time.[/quote]

It's a sad state of affairs when someone is fearful of being reproached for their opinion in a comment field who clearly is NOT a troll

 

[wiyosaya]

IMHO, this is very encouraging news. It sounds like AMD has realized that you need someone with creative genius (rather than a competitive emulator) designing your chips.

According to Wikipedia, K8 brought AMD64 to market, and put the memory controller on die. Personally, I think these two innovations were largely responsible for AMD taking the crown away from Intel in those days.

Arguably, AMD64 was the way to go for 64-bit rather than Intel's take a left turn at Albuquerque approach - Itanium (that some liked to call ITanic). AMD64 was a huge win for AMD as far as influencing what was at the time future CPU technology.

As well, having the memory controller on-die arguably substantially improved memory access times while simplifying over-all system design.

The article does not make it clear if Keller was behind those innovations, but I would expect that he had at least some input into the design. The question in my mind is whether Keller will be able to be creative enough to take what they have, the much maligned Bulldozer, and fine-tune it through perhaps some architecture changes or to take an ever larger step and come up with a substantially improved architecture that might be completely different from BD. The former should be cheaper than the latter though not necessarily better, and with near-sighted, blurry vision business climates these days, short-term cost considerations usually out-weigh more expensive approaches that may be significantly more successful.

As I see it, AMD needs something like the technological steps that K7/K8 brought to market to once again become competitive with Intel. With this announcement, I am encouraged that it sounds like someone finally convinced the new CEO that if AMD fights, something fruitful is likely to come from it.

Time will tell. I've been an AMD fan for some time, however, after disappointing reviews of BD, and the simple fact that the I7-3820 was a significantly better performer for the types of apps that I run, I went with Intel this time around. If BD had been close to the 3820 in running the kinds of apps that I run, I would not have hesitated to build with BD. However, the choice between the 3820 and BD seemed like a no-brainer to me.