AMD has appointed a new corporate VP, Nazar Zaidi, to head future development of CPUs, SoCs, and systems engineering. While Zaidi isn’t being billed as a direct replacement for Jim Keller, AMD’s former CPU architect, his position seems to blend both engineering and long-term product focus and development. Keller left the company last summer after completing the Zen design; the CPU design team has been headed by Chief Technology Officer (CTO) Mark Papermaster during the interim period.
Nazar Zaidi was formerly a VP of engineering at Broadcom and worked on that company’s custom ARMv8 processor, dubbed Project Vulcan. That project appears to have quietly come to an end after Broadcom was acquired by Avago a year ago — at the very least, there’s been no significant news for nearly a year. Prior to working for Broadcom, he was the VP of engineering for NetLogic and oversaw development of the XLP980, a high-end networking and communications processor (NetLogic was acquired by Broadcom).
Jump back to the 1990s, and Zaidi led development of several aspects of the original Itanium, including its bus architecture and hardware compatibility implementation. The performance of said hardware implementation was considered poor at the time. I’ve never seen a breakdown of why Merced’s x86 performance was so awful, but the chip suffered numerous delays and development issues. To some in the CPU community, working on Itanium might be considered a negative rather than a strength — I think such a viewpoint is rather premature.
A difficult road
With that said, I think it would be a mistake to paper over the scope of the challenge facing both Zaidi and AMD itself. Even if Zen delivers on all of AMD’s promises, the new CPU core isn’t going to close the gap between AMD and its rival, Intel, in a single bound. APU hardware still needs to be brought to market, AMD will need to decide when and if it wants to develop technologies like HBM2 for the desktop and laptop space, and the PC market remains soft.
Lisa Su has laid out plans for AMD to regain ground in server and mobile as a way to offset declining PC revenues. While the PC market as a whole may be declining, any win that AMD can eke out that increases its overall market share can still drive better revenue and improve products. Even if Zen is a great CPU, it’s the first in what needs to be a fundamental series of new products that give AMD what it needs to reposition itself. To put this somewhat differently: What AMD needs to do with Zen is analogous to what it did with the K7 — except it’s going to be launching this part into a much tougher market with PC sales slumping and the performance gap between itself and Intel currently larger than it was during the K6 / P2 era.