AMD Ryzen AI Max+ Outperforms Nvidia RTX 4090, Apple M4 Pro, and Intel Ultra 9 in Windows

AMD Unveils Powerful Ryzen Processors at CES 2025

During the CES 2025 event, AMD introduced its cutting-edge Ryzen 9 9955HX3D mobile processor. This new chip features a 3D V-cache architecture and joins the recently released Ryzen 9950X3D and Ryzen 9900X3D desktop processors. Additionally, the company launched the Ryzen 200 series CPUs, targeting the mainstream market.

Expanding the Ryzen AI Lineup

AMD has also broadened its Ryzen AI 300 series APU offerings, which previously included the Ryzen AI 9 HX 375, 370, and 365. New additions consist of more cost-effective options: the Ryzen AI 7 (PRO) 350 and Ryzen AI 5 (PRO) 340. Alongside these, AMD is launching the more potent Ryzen AI Max and Max+ models.

It’s important to note that the newly released Ryzen 200 series chips do not meet the performance threshold in this new classification system, as they peak at just 16 TOPS.

Classifying AI PC Chips

Moving forward, AMD aims to categorize its AI processors into three distinct classes: Halo, Premium, and Advanced. In this hierarchy:

Max and Max+ will represent the Halo class;
Ryzen AI 9 will slot into the Premium tier;
Ryzen AI 7 and 5 will comprise the Advanced category.

Performance Claims and Comparisons

The new Ryzen AI Max+ parts are positioned as performance leaders, with AMD asserting superiority over competitors such as the Nvidia RTX 4090, Apple M4 Pro, and Intel Ultra 9 288V. For instance, the Ryzen AI Max+ 395 APU reportedly outperforms Intel’s 288V by an average of 2.6 times in rendering tasks and offers about 1.4 times the speed in synthetic gaming scenarios.

When compared to Apple’s 14-core M4 Pro, AMD anticipates improved rendering performance with applications such as Blender, Corona, and V-Ray, with V-Ray yielding the most significant performance advantages.

Moreover, AMD’s Max+ 395 APU is expected to excel in AI processing tasks, claiming it is 2.2 times faster than the Nvidia RTX 4090 in terms of tokens processed per second on LM Studio. AMD refers to the 395 as the “World’s first Copilot PC+ processor suitable for running 70 billion LLM.

Innovative Memory Architecture

A key feature that enhances the capabilities of the Max APUs is the ability to allocate up to 96 GB of memory to its integrated 40 CU RDNA 3.5 graphics, out of a total of 128 GB available. The power behind these new Max and Max+ processors stems from their architecture, which boasts 16 Zen 5 cores alongside a 50 TOPS capable XDNA 2 NPU.

Another fascinating advancement is the introduction of the “Unified Coherent Memory Architecture”, a new memory interface that provides a staggering 256 GB/s memory bandwidth. This innovation further sets the Ryzen AI Max and Max+ APUs apart from their predecessors.

Competitive Insights

In addition to the Max+ performance metrics, AMD highlighted its expectations for the Ryzen AI 7 350 to outperform both Intel and Qualcomm. The company estimates the 350 APU will achieve performance up to 49% better than the Snapdragon X Plus X1P-42-100 and up to 78% faster than the Intel Core Ultra 7 258V.

To explore the complete specifications and availability details for the Ryzen AI 350 and 340, refer to the image above.

Source & Images