AMD Ryzen AI Max APUs: Performance Insights and AI Acceleration in Mobile Chipsets

AMD Ryzen AI Max Geekbench scores reveal a power drop in 300-series APUs

Not all laptop chips are created equal, and that's as true of AMD's innovative Ryzen AI Max "Strix Halo" APU as any. AMD unveiled the Strix Halo APU in January as the Ryzen AI Max 300 series, debuting three consumer chipset variants and four variants made for workstation-class machines. While we normally expect a large performance gap in between, say, a Ryzen 5 and Ryzen 9. But, with a name like Ryzen AI Max, you may not be expecting such a performance gap between the three APUs in the AI Max series. An APU, or accelerated processing unit, is a chipset that combines the CPU and an integrated graphics tile. AMD coined the term back in 2011. AMD uses APU for all of its mobile chipsets, from the Ryzen AI Max to the Ryzen AI 300 series. So far, we've seen only systems powered by the 16-core Ryzen AI Max+ 395 chip, including the Asus ROG Flow Z 13 gaming laptop. However, the Ryzen AI Max family offers more budget-friendly options, too, from the 12-core Ryzen AI Max 390 down to the 8-core Ryzen AI Max 385. Last week, we finally saw the 8-core Strix Halo chipset's performance. An HP ZBook Ultra 14 G1a featuring this budget-friendly APU has been benchmarked on Geekbench 6, and the results were uploaded to the Geekbench archives. So, how does the 8-core Ryzen AI Max PRO 385 stack up against the 16-core 395? Let's take a look. According to results uploaded to Geekbench on May 27, the HP ZBook Ultra 14 G1a with an AMD Ryzen AI Max PRO 385 APU scored a Geekbench single-core score of 2,489 and a multicore score of 14,136. We expected these scores to be higher. However, a few factors may explain the larger-than-expected performance gap. First, there's the hardware matchup (8 cores vs. 16). The Ryzen AI Max PRO 385 also has a lower max frequency than the flagship, which could explain the dip in single-core performance. The difference in RAM between the HP ZBook Ultra configuration Laptop reviewed and the ZBook scores uploaded to Geekbench could also be behind the score differences, as RAM affects how Geekbench scores are calculated. While the Ryzen AI Max 385 and its professional variant will still be interesting chipsets -- as both still feature the larger Radeon 8060S integrated graphics tile -- these early benchmarks clarify a few things. Based on AMD's naming convention, the Ryzen AI Max 300 is designed to sit directly atop the Ryzen AI 300 series. So, as the entry-level Ryzen AI Max chip, the 385's Geekbench scores are just above what we've seen from the top end of the Ryzen AI 300 series. We're not certain yet how expensive the Ryzen AI Max 385 will be compared to its slightly less powerful Ryzen AI 300 series counterparts, as only two Ryzen AI Max systems are on the market so far. The HP ZBook Ultra with the Ryzen AI Max 385 starts at $2,599. Meanwhile, the Asus ROG Flow Z13 doesn't have a Ryzen AI Max 385 variant but costs $2,099 for the slightly more powerful Ryzen AI Max 390 chipset. But based on those prices, you are paying quite a bit more for the Ryzen AI Max chipset and its more powerful Radeon 8060S iGPU.

AMD Ryzen AI Max Pro 385 Geekbench Benchmark Leak Reveals Performance

The AMD Ryzen AI Max Pro 385 APU's Geekbench results have been published, revealing its performance profile under an HP ZBook Ultra G1a testing environment. The APU was evaluated using the system's "Optimized" power mode, which can influence thermal headroom and overall throughput. In single-core testing, it earned a score of 2,489 points; in multi-core testing, it reached 14,136 points. These figures provide a reference point for understanding its raw computational capacity. For context, the Ryzen AI Max Pro 390 APU was benchmarked on a similar but slightly different platform: the HP ZBook Ultra Mini G1a workstation. That component recorded a 2,759-point score in the single-core test and 15,779 points in the multi-core test. When comparing these results, the Pro 385's performance is approximately 10 percent below the Pro 390's scores in both categories. It is worth noting that variations in system cooling, BIOS configurations, and power profiles can affect benchmark outcomes. Consequently, while the Pro 385 shows solid results, external factors such as chassis design and firmware can impact final numbers in real-world applications. The architecture is based on eight Zen 5 CPU cores, each running at a base frequency of 3.60 GHz, with the ability to boost up to 5.00 GHz under load. On the graphics side, it integrates a Radeon 8050S GPU that delivers 32 compute units dedicated to accelerating graphics and parallel compute tasks. Additionally, the APU incorporates a neural processing unit rated at 50 TOPS, enabling faster inference for AI workloads such as neural network model execution, computer vision tasks, and machine learning preprocessing. This heterogeneous design is intended to balance general-purpose CPU performance, GPU-accelerated compute, and specialized AI acceleration within a mobile power envelope. Given these specifications, it becomes clear why the Ryzen AI Max Pro 385 is positioned just below the Pro 390 in AMD's line of mobile workstation APUs. The Pro 390 likely offers higher clock frequencies, possibly more GPU execution units or NPU resources, which translate into the roughly 10 percent performance uplift observed in Geekbench. For professionals requiring AI inference acceleration on the go, the Pro 385's 50 TOPS NPU and 32 CU GPU can still handle demanding workloads while consuming less power than the higher-tier models. Source: geekbench