AMD announced its Ryzen 8000 processors last week, codenamed “Hawk Point,” highlighting the integrated NPU (Neural Processing Unit) for AI workloads. These chips are (mostly) the same as the Ryzen 7000 family, leveraging the Zen 4 CPU and the RDNA 3 graphics architecture. The NPU leverages the XDNA architecture to deliver up to 39 TOPS of processing power. Hawk Point is a refresh of Phoenix Point with the same CPU/GPU combo, but a beefed-up NPU.
The only tangible difference is the AI performance, most notably generative AI models. According to AMD’s official figures, the Ryzen 8040 processors are up to 40% faster than the preceding 7040 chips in generative AI. In this case, the Ryzen 9 7940HS is compared to the Ryzen 7 8840HS (a lower trier part) in Llama 2 and Vision Models.
The Ryzen 8040 processors boast higher gaming, and content creation performance than Intel’s 13th Gen Raptor Lake-H processors. Official first-party benchmarks place the Ryzen 9 8940HS as 80% faster than the Core i9-13900H in gaming and 40% faster in content creation.
Unfortunately, since the CPU and GPU specs are identical to the Ryzen 7040 processors the non-AI performance should be too. We are looking at a pure rebrand to dampen the Meteor Lake launch set to happen in the coming days.
AMD also provided an update on its Ryzen mobile roadmap. The Ryzen 8040 “Hawk Point” family is launching in the first quarter of 2024 (shipping now).
Strix Point is next in line with a shipping date of “2024.” It is expected to feature a hybrid core CPU with Zen 5 and Zen 5c cores, RDNA 3.5 graphics, and the XDNA 2 NPU. The company claims that 2nd Gen Ryzen AI will be over 3x faster than existing solutions in generative AI.