AMD Medusa Point debuts on Geekbench with 10-core CPU

The first signs of AMD's next-generation “Medusa Point” mobile silicon have surfaced on Geekbench, providing a first look at the architecture expected to arrive in laptops next year. While the benchmark scores are negligible due to the chip running at a low clock speed of roughly 1.39GHz, the technical specifications revealed in the entry confirm major architectural shifts.

The chip is listed on Geekbench (via Wccftech) as a 10-core, 20-thread part, similar to the Ryzen AI 9 365 and Ryzen AI 9 465. Given that all cores appear to have SMT (two threads per core), none of them should be low-power cores dedicated to background tasks.

Perhaps the most significant piece of information taken from the Geekbench entry is the cache upgrade. The sample features 32MB of L3 cache, a 50% increase over current 10-core parts like the Ryzen AI 9 365 (24MB) and double that of the older Hawk Point (16MB). Each core also has 1MB of L2 cache. Medusa Point is rumoured also to feature a new type of integrated graphics, combining RDNA 5 and RDNA 3.5 IP with an updated NPU to meet the increasingly demanding local AI requirements of 2027-era operating systems.

While a launch isn't expected until 2027, the presence of a functional engineering sample in a public database suggests that AMD is well into the validation phase of the 3nm Zen 6 process.

KitGuru says: The increased cache is a nice upgrade but shouldn't be the only one on the CPU side of this APU. What other specs do you expect to see improve with the new generation of Ryzen AI mobile APUs?

Become a Patron!