As tensions begin to mount over “AI” chips with competitive iGPUs, in particular for use in Copilot PCs , we’ve already been able to gain some insight into the two very different strategies being pursued by AMD and Intel. While Intel seems to have a promising solution in Lunar Lake, AMD is exploring a different path and is said to have a few more surprises up its sleeve, with versions superior to the already known Ryzen AI 370 and 375. AMD’s aim is to meet the growing demand for more robust AI performance, particularly in high-end notebooks and workstations.
Ryzen AI 300: Strix Point and Strix Halo in the same range?
Rumors point to larger and more complex models than those currently available, suggesting a revised architecture optimized for AI and HPC tasks. The addition of more powerful processing cores and increased memory bandwidth should enable these new APUs to handle heavier workloads, while maintaining optimized energy efficiency.
However, this increase in performance comes at a price: a higher TDP. The source of this information is considered reliable in view of his regular confirmed information on AMD. We’ve known for some months now that AMD is targeting a high-end APU to compete with Apple’s M SoCs used in desktop versions.
One hypothesis would be to see the arrival of the Strix Halo APU within the Ryzen AI 300 range. This APU would be more robust, particularly in AI processing, but would also have a larger packaging. Its TDP would also be higher, and it could be used in a customizable way by AMD’s partners. Up to now, we’ve been talking about a Mini PC / workstation target for the Strix Halo. In its top configuration, this APU would feature 16 Zen 5 cores, 40 RDNA 3.5 processing units and a larger L3 cache with 32 MB per CCD. For the moment, the release date of this APU is uncertain, with the latest rumors suggesting a first presentation during CES 2025 (January).
