Ex-AMD, Apple CPU Architect Keller Heads to Tesla

The key architect behind some of the more successful processors for Advanced Micro Devices and Apple will now head up the self-driving automobile efforts at electric car maker Tesla.

Jim Keller, who left AMD in September 2015 after his second stint at the chip vendor, will be vice president of Tesla’s Autopilot Hardware Engineering unit, according to the car company. According to a statement put out by Tesla officials, “Jim will bring together the best internal and external hardware technologies to develop the safest, most advanced autopilot systems in the world.”

Self-driving cars require a lot of processing power, which fuels the need for greater energy efficiency in their designs. Tesla currently uses Nvidia’s ARM-based Tesla systems-on-a-chip (SoCs) for its in-car systems, and the chip maker has made such emerging market segments as machine learning and self-driving cars a priority for its SoCs and its GPUs.

A growing number of tech companies and automakers are putting a lot of effort behind self-driving cars.

Keller was a driving force behind some of AMD’s most successful efforts during his two tenures with the company. In the late 1990s, he helped develop the K7 and K8 Athlon architectures and the 64-bit x86 chips, which were introduced with the company’s first Opteron server processors in 2003 and helped grow AMD’s server market share to more than 20 percent. Opteron also forced Intel to push aside its Itanium initiative and bring 64-bit capabilities to its own x86-based Xeon chips.

He left AMD for semiconductor maker P.A. Semi, which was later bought by Apple. While at Apple, he worked on the device maker’s A4 and A5 chips for iPhones and iPads. He returned to AMD in 2012, leading the development of the “Zen” CPU, a core design more than two years in the making that AMD officials say will play a key role in its future PC and server chips. It will support simultaneous multi-threading (SMT)—a technology similar to Intel’s Hyper-Threading—and DDR4 memory, and will feature a FinFET transistor design for the 14-nanometer chips.

The Zen architecture will first appear in chips for desktop PCs later this year, and will ramp in servers in 2017, according to officials. Engineers at AMD already are working on the follow-on to Zen, which officials last year called “Zen+.” While losing Keller was a blow to AMD, officials said the bulk of the work for Zen had been completed by the time he left.

AMD officials are banking on Zen to help the company become more competitive with larger rival Intel in the high-end PC and server markets.