Notebookcheck Logo

AMD outlines plans to integrate AI XDNA IPUs across its entire processor portfolio

AMD desktop and server-grade AI-powered CPUs could be a thing in a few years. (Image Source: Freepik)
AMD desktop and server-grade AI-powered CPUs could be a thing in a few years. (Image Source: Freepik)
For now, only select mobile-grade processors will feature the AI XDNA IPUs. AMD is considering IPUs for desktop and server CPUs some time in the next three years, not before software support sees clear expansion.

In a recent PCWorld interview, AMD's David McAfee - DVP & GM of Client Channel Business outlined Team Red’s plans to expand the integration of AI Inference Processing Units based on Xilinx’s IP in upcoming mobile, desktop and even server-grade processors. Right now, only the “Phoenix” Ryzen 7040U and 7040HS processors pack the AI “XDNA” IPUs, but AMD is now carefully considering what other processor classes could take advantage of AI-powered hardware.

For mobile processors, the IPU workload would focus for example on video communication features like keeping the camera focus on the speaker, maintaining eye contact as much as possible and providing high quality background effects, as well as voice translation whenever necessary. For gaming desktops built with performance in mind, the IPUs would improve the existing “SenseMI” technology, adjusting CPU clock speeds using Precision Boost, mXFR and Smart Prefetch in a more efficient way. HPC Threadrippers and server-grade EPYC processors could also integrate IPUs, making AI training more streamlined, but such powerful CPUs would not necessarily benefit from AI otherwise.

However, the integration costs and the overall value brought for desktop and server processors is not yet justified. Even if mobile-grade IPUs are easier to implement because of their superior power efficiency, AMD needs to see a “more concrete” end-user benefit before adding AI to more low-end solutions.

The integration of AI across all processor classes would therefore depend on software advancements. For now, much of the AI hype revolves around language models and generative models (text-to-image etc) and these do not really need CPU-based inference since the models are mostly trained in the cloud with powerful GPUs. We first need to see more practical implementations like the OS-level integration rumored for the upcoming Windows 12.

In this regard, McAffe concludes by estimating that “over the next three years, the software and user experiences have to deliver that value, and move this from a really exciting, you know, sort of emergent technology that’s just making its way into the conversation on PCs, into something that it potentially is rather transformational to the way that we think about performance, and devices, and how we use them.”

 

Buy the OBSBOT Meet 4K AI-Powered Webcam on Amazon

Source(s)

Read all 3 comments / answer
static version load dynamic
Loading Comments
Comment on this article
Please share our article, every link counts!
> Expert Reviews and News on Laptops, Smartphones and Tech Innovations > News > News Archive > Newsarchive 2023 05 > AMD outlines plans to integrate AI XDNA IPUs across its entire processor portfolio
Bogdan Solca, 2023-05-16 (Update: 2023-05-16)