AMD is considering deeper integration of AI technology in the Ryzen product line
At the commencement of this year’s CES 2023, AMD unveiled the Ryzen 7040 series APU, one of its highlights being the incorporation of an AI acceleration engine based on the XDNA architecture. This inclusion has notably diminished the utilization of x86 cores and CUs, bearing similarity to Apple’s M2 series chip’s AI engine. AMD intends to leverage the AI acceleration engine for various AI-accelerated tasks, such as image processing, with plans to introduce a standalone SDK for developers.
David McAfee, AMD’s Vice President and General Manager of the Customer Channel, recently partook in an interview with PC World. He broached the topic of the AI acceleration engine within the XDNA architecture, expounded on current AI hot topics, and elucidated the potential impact on AMD’s future product line.
ASUS ROG Custom Ryzen Z1 APU vs. AMD Phoenix, Source: Engadget/AMD
With the inclusion of an AI acceleration engine inevitably leading to additional costs, AMD is assessing its value via the Ryzen 7040 series APU. Before extending to other chips, the benefits users garner must be more tangible. AMD envisions AI ascending to a position of equal importance with the CPU and GPU in the future, becoming a third pillar of computing that offers greater value to the platform. However, currently, there is no robust industry standard or benchmark to assist users in understanding, and even application developers are still striving to discern what AI can accomplish.
The decision to introduce an AI acceleration engine first in mobile platform chips is primarily driven by a desire for energy-efficient computation when frequent calculations of a particular type are required. However, for the desktop platform, which is less concerned with power efficiency, this incentive is relatively weaker. According to McAfee, one of the most crucial challenges to be addressed before the Ryzen product line can integrate AI technology more deeply is the support of client applications and operating systems.
There are two reasons for choosing to run AI applications on personal computers rather than in the cloud: one is that local acceleration on mobile platforms can yield benefits for battery life; the other is to ensure security, eliminating reliance on the cloud and thereby mitigating potential data leakage. With Microsoft intending to integrate AI more closely with Windows 12, McAfee believes this to be critical in conveying the tangible value of AI as an emerging technology to developers and users.
