Jim Keller has left AMD. So? Do we have to to rip the clothes off? Do we have to think that AMD is almost dead? Do we have to think that Zen will be a failure? No, we don't.

 

 

Jim Keller left AMD during 1999, after just one year at Sunnyvale. There, during 1 (ONE!) year, he created with his team the HyperTransport technology and the x86-64 ISA (Jim Keller was lead architect of the AMD K8 microarchitecture). He was the father of the K8, and the K8 was commercialized just during 2003. Why? Because of some problems with the foundry's technology. K8 was almost completed during 2000.

Today, after 3 years at AMD, Jim Keller left Sunnyvale again. Zen and Zen+ are completed, and AMD has to put this microarchitecture on a wafer. The work of Jim Keller is finished here.

So, I think that the trolls, the fanboys and the analysts on drugs have to shut up.

I have a personal idea why Keller went back at AMD. I wrote about this at S|A forum, and at our forum (in Italian), 4 months ago.

Keller is now 56 years old. When Zen will be in the store, Keller will be 57 years old. I think that he joined AMD to do something that never be forgotten. Money was his last troubles. In Intel and Apple engineers have to follow the instruction of the executives. In AMD Keller can do whatever he wants, he's the boss.

"Mr. Keller, do you want create a SMT arch? Sure! Do you want to do an ARM microarchitecture? Sure! Do you want to trash Bulldozer? Sure! Do you want to redesign the Cache of our CPUs? Sure!", etc. Jerry Sanders, first CEO of AMD, said: "People first, products and profit will follow!". IMO, he was right.

So, by the point of view of an engineer, what is the best thing to do, to be remembered forever and ever? Kick the Intel ass, with a powerful microarchitecture created thanks to a stick, some paper clips and two penny (AKA, the AMD R&D budget).