AMD Zen 6 Laptop CPU Reaches 3,329 Points in Early Geekbench Test

An early engineering sample of an AMD Zen 6 mobile processor has appeared in a new Geekbench result, showing a notable improvement over the company’s current laptop chips.

The processor carries the identifier AMD Eng Sample 100 000001713 33 N and runs on a development platform called Plum MDS1. It has 10 CPU cores, 10MB of L2 cache and 32MB of L3 cache.

The sample recorded 3,329 points in Geekbench 6.5.0’s single core test and 16,555 points in the multi core test. These results are considerably higher than those produced by an earlier sample carrying the 31 N designation.

The new identifier may represent a more recent processor revision, updated firmware or an improved development platform. The basic specifications appear unchanged, suggesting that AMD may be improving performance as Zen 6 moves closer to production.

Specification	Early Zen 6 sample
Development platform	Plum MDS1
Engineering sample	100 000001713 33 N
CPU cores	10
Core configuration	Reportedly 4 plus 6
L2 cache	10MB
L3 cache	32MB
Single core score	3,329
Multi core score	16,555
Expected product family	Medusa Point mobile processors

Single core performance is the main highlight

The single core result is the strongest part of the benchmark. A score of 3,329 places the engineering sample above several powerful laptop processors, including models designed for large gaming and workstation notebooks.

It also represents a major improvement over the Ryzen AI 9 HX 370, one of AMD’s leading current generation mobile processors. That chip has a typical Geekbench single core score of around 2,606, although results can change depending on the laptop, cooling system, memory and power settings.

Using those figures, the Zen 6 sample is approximately 28 percent faster in the single core test. Comparisons against independently tested Ryzen AI 9 HX 370 laptops produce a smaller improvement, but the new sample still maintains a meaningful advantage.

Single core performance affects many common activities, including application responsiveness, web browsing, game logic and tasks that cannot use a large number of CPU cores. A strong result would therefore be important for both thin laptops and larger performance focused systems.

The score should still be treated carefully. Engineering samples often run with unfinished firmware, unusual power settings and development drivers. Geekbench results can also vary between individual test runs.

Multi core performance may be limited by power

The sample’s multi core score of 16,555 is less dramatic when viewed on its own, but the processor only has 10 cores. It reportedly combines four larger performance focused cores with six smaller dense cores.

Compared with the Ryzen AI 9 HX 370’s average score of approximately 13,400, the Zen 6 sample is around 24 percent faster. This is notable because the Ryzen AI 9 HX 370 has 12 CPU cores.

The result suggests that Zen 6 could deliver higher performance from each core, allowing a 10 core processor to compete with or outperform current 12 core designs.

Power limits may have prevented the sample from reaching its full multi core potential. Laptop processors must remain within strict thermal and electrical limits, especially when tested on early development boards. The final performance will depend on how manufacturers configure cooling and sustained power consumption.

The chip may be an early Medusa Point processor

The Plum MDS1 platform is believed to be associated with Medusa Point, AMD’s future mobile processor family based on Zen 6 architecture.

Medusa Point is expected to use more than one type of Zen 6 core. The reported four plus six arrangement may combine four full performance cores with six compact cores designed to provide greater efficiency and core density.

Both types are expected to support the same basic instruction set, but they may use different internal layouts, clock speeds or cache arrangements. This would allow AMD to balance responsive performance with battery life and multi core capability.

Clock speeds were not clearly established in the benchmark, making it difficult to determine how much of the gain comes from architectural improvements and how much comes from operating frequency.

Final retail processors may perform better as AMD improves firmware, memory support and power management. They could also perform differently if commercial laptops apply more conservative thermal limits.

The result offers an encouraging early sign for Zen 6, particularly in single core workloads. However, one benchmark cannot confirm overall performance. Gaming tests, productivity applications, power measurements and sustained workloads will be needed before the architecture can be compared properly with AMD’s current processors or Intel’s upcoming Nova Lake family.