AMD CTO: Hybrid cores are coming soon to Ryzen processors and generative AI will speed up chip design
Mainstream hybrid processor cores have been around for around a decade now. First introduced by Arm with its big.LITTLE architecture for smartphone chips, this design combining bigger performance cores with smaller efficiency cores made its way into laptop processors (Apple M / Intel Tiger Lake) and eventually desktop processors (starting with Intel Alder Lake) in the past few years. Despite using the hybrid approach, Intel’s desktop processors are still quite power hungry compared to AMD’s current Zen 4 core architecture that does not rely on any hybrid designs. Apple’s M chips, on the other hand, are considered some of the most power efficient on the market. Team Red is somewhere in the middle right now, but this hierarchy could soon change, as AMD is already planning to add its own spin on the hybrid core design assisted by generative AI.
In a recent Tom’s Hardware interview at the ITF World event, AMD’s CTO Mark Papermaster talks about Team Red’s upcoming hybrid core design for desktop processors:
[...] You'll see high-performance cores mixed with power-efficient cores mixed with acceleration. [...] It's not only how you've optimized for either performance or energy efficiency, but stacked cache for applications that can take advantage of it, and accelerators that you put around it.
This hybrid design will further expand to server-grade processors and the integration with AI applications will play an important role:
As AI moves from not only being in the cloud, where the heavy training and large language model inferencing will continue, but you're going to see AI applications in the edge. And it's going to be in enterprise data centers as well. They're also going to need different core counts and accelerators.
AMD plans to use AI in order to come up with more efficient chip designs and speed up the verification / validation process:
So we are applying AI today in chip design. We're using it in 'place and route,' both in how we position and optimize our sub-blocks of each of our chip designs to get more performance and to lower the energy. [...] It's looking at what patterns created the most optimal design, and so it's actually speeding the rate of having an optimized layout of your chip design elements, and therefore giving you higher performance and lower energy, just like we're doing with how we optimize our chiplet partitioning and chiplet placement.
Furthermore, generative AI could eventually evolve to design entire microarchitectures, but it will not completely replace designers:
There's no question that we're just at such an early phase where you can start thinking about generative AI to actually create new approaches and build upon existing designs you have. [...] It won't replace designers, but I think it has a tremendous capability to speed design.