Just when you thought you were away from more AI-centric hardware news, AMD is entering the “AI PC” space with its latest mobile notebook chip. The new Ryzen AI 300 series has better performance than Intel or Qualcomm, in addition neural processing power.
The chip industry has always been a game of competition. Now, chipmakers are working harder than ever to compare their CPUs and GPUs not just on power, but on the promise of ultimate PC performance in the future brought about by the spread of artificial intelligence. AMD doesn’t have to compete with long-time rival Intel in the consumer PC market, but Qualcomm, largely thanks to ARM-based Snapdragon X Elite and X Plus in the latest Co-pilot+computer.
AMD is mainly focused on promoting its two new chip families. One is a new version of Ryzen CPUs, including the Ryzen 9000 series, and the other is the Ryzen AI 300 series, which comes with a new NPU in the form of XDNA 2. AI9HX370.
Microsoft says it needs NPUs with at least 40 TOPS to label them for use on Copilot+ PCs. Like recent Snapdragon chips, the HX 370 and 365 have the same NPU, running at 50 TOPS. This was one of the biggest boasts about AI performance last year, but while the company claims it can run more complex AI models, we still need to see if any software is worthy of these new neural components.
The 370 has 12 cores, 24 threads, and a maximum boost speed of 5.1 GHz, while the 365 has 10 cores, 20 threads, and a maximum speed of 5.0 GHz. The chip also has an RDNA 3.5 built-in GPU, suitable for some mobile graphics work or gaming. The chips will appear in new laptops in the coming days during Computex Taiwan.
What is Zen 5?
These new chips use AMD’s new centralized architecture, known as Zen 5, on the CPU side. The chipmaker claims Zen 5 is a major update Compared to Zen 4, which should handle twice the bandwidth of the previous generation. What does this mean for PCs? AMD promises you’ll see up to a 19% improvement in benchmark performance in Geekbench 6, or a 13% improvement in the physical 3DMark test, but that will depend on your PC’s specific chip and other architecture.
Zen 5 and XDNA 2 NPU architecture are different. You can break down TOPS speeds in an NPU into a bunch of other categories, but AMD claims XDNA 2 is twice as energy efficient as the previous generation at 10 or 16 TOPS, and has many times the total neural computing power of the previous generation.
AMD’s new gaming CPU improves performance with Zen 5 architecture
For those who don’t care much about AI 300 chip-centric productivity machines, all you want to know is how the Ryzen 9000 series compares to the previous generation and Intel’s latest offerings. These include Ryzen 5 9600X, Ryzen 7 9700X, Ryzen 9 9900X, and the highest-end is Ryzen 9 9950X. Most have slightly higher clock speeds, but some (like the 9900x and 9700X) are more energy efficient and have better TDP.
To be honest, the 9950X has 16 cores, 32 threads, and clock speeds of up to 5.7 GHz. Technically, these are the same specs as the previous generation Ryzen 9 7950X3D. AMD is trying to challenge the Intel Core i9-14900K, claiming you’ll see slightly better frame rates in games such as Cyberpunk 2077 or F1 2023 Thanks to the new Zen 5 architecture, the bandwidth for multitasking is much greater.
All these gaming-centric CPUs should be available this July. This is great news for anyone with a motherboard that supports the AM5 socket. AMD has committed to supporting AM5 until 2027, so if you’re considering upgrading, you’ll have the opportunity in the next few years. After more than eight years of operation, AMD plans to end support for AM4 in 2025 or sometime thereafter.
The chipmaker said pricing for the 9000-series chips has yet to be determined, but we should know more when they launch in July.
