AMD Ryzen processors are currently beyond doubt, being very good quality chips. During a conference, Jim Keller talked about the K12 ARM project, AMD processors based on the ARM architecture. It has been highlighted that AMD canceled the project “foolishly” after Keller himself left the company.
It is possible that the name of Jim Keller don’t tell yourself much for the moment. To put things into context, it’s the father of zen architecture 1in addition to laying the groundwork for Zen 2 and Zen 3. After leaving AMD, I worked for a few years at Intel, being one of the architects of the design of the power and efficiency cores of the current Core Alder Lake.
Keller blames AMD’s decision to cancel ARM processors
During a conference called “Future of Compute” organized by the Department of Computing and Automation of the Instituto Indicio de Ciencias, he spoke about the different projects he has worked on. Keller He talked about the projects he worked on and the fundamentals of chip design.
At AMD, Keller helped develop the Zen 1 architecture.. Additionally, Keller laid the groundwork for the Zen 2 and Zen 3 architectures. After the current Zen 3 architecture (Ryzen 5000), the Zen 4 and Zen 5 designs will no longer be based on Keller’s ideas, so to speak.
Jim notes that when he worked at AMD, he and his team they noticed that the cache layout between ARM and x86 processors was about the same. They found that the execution unit was practically the same. They discovered that the only difference between the two architectures was in the decoding unit.
Faced with this discovery, they decided to work on the development of a new chip. This new the project was named K12
Not only that, he says, managers are often afraid of change. He points out that since he was a (processor) architect he was not afraid of changes, like this one. Additionally, Keller says the work done while at AMD was “fun”
The K12 chip was based on the ARMv8-A CPU and it was planned to launch in parallel with the Zen architecture. This chip was focused on high-workload, high-power-efficiency environments. Specifically designed for high density servers, embedded systems and semi-custom solutions.
AMD Response
Unsurprisingly, AMD reacted to this information, as Keller is a major figure in the industry. Who made the comment is AMD CFO Devinder Kumar. Comment the following:
But I’ll tell you from my point of view, when you look at computing solutions, whether it’s x86 or ARM or even other areas, that’s an area that we’re focused on investing in for us. We know how to calculate very well. Even ARM, as you mentioned, we have a very good relationship with ARM. And we understand that our customers want to work with us on that particular product to provide the solutions. We are ready to go ahead and do it even if it’s not x86, even if we think x86 is a dominant force in this field.
Perhaps, and with all due respect, someone from the economics division is not the best person to respond to this comment. AMD has engineers who at any time could be more valuable. With this answer, it seems (at least from my perspective) that it was only canceled because it was not financially viable.