HyperAI
Back to Headlines

AMD's Zen 6 "Medusa Point" APUs May Boast Up to 22 Cores, Leaks Suggest

16 hours ago

According to recent rumors, AMD's next-generation Zen 6 "Medusa Point" APUs could come equipped with up to 22 cores, a significant leap in performance and efficiency. Leaker HXL, supported by several other industry sources, claims that these APUs are designed with a mix of classic, dense, and low-power cores, all based on the upcoming Zen 6 architecture. While the launch of Zen 6 is still at least a year away, and the mobile variants may not debut until early 2027, it's important to approach this information with caution. "Medusa Point" is expected to succeed AMD's current Zen 5-based "Strix Point" APU series. However, rather than a direct transition, AMD is reportedly developing a transitional lineup called "Gorgon Point," which will serve as a refreshed version of Strix Point. This interim step suggests that AMD is methodically preparing for the full Zen 6 rollout. Architecturally, Medusa Point will shift to the Zen 6 core design, which AMD is expected to unveil around Computex next year. The graphics engine is rumored to incorporate the RDNA 3.5+ architecture, although many had hoped for RDNA 4. It seems that RDNA 4 may be reserved for standalone GPUs, with the more advanced UDNA 1 or RDNA 5 expected to launch simultaneously with Medusa Point. For the mainstream Ryzen 5 and Ryzen 7 models within the Medusa Point series, up to 10 hybrid cores are predicted. This configuration includes four classic Zen 6 cores, four dense Zen 6c cores, and two low-power (LP) cores. The LP cores are likely smaller and optimized for maximum energy efficiency, making them particularly suitable for mobile devices. The integrated GPU (iGPU) is anticipated to have eight compute units (CUs) based on the RDNA 3.5+ design, similar to the Radeon 860M. This represents a step back from the current 16-CU design found in the Radeon 890M, but it was probably a trade-off to allocate more space to other critical components. The RDNA 3.5+ architecture is essentially the second refresh of the RDNA 3 line, and it is expected to bring enhancements in machine learning capabilities. These improvements could pave the way for the fourth generation of AMD's FidelityFX Super Resolution (FSR 4). The mainstream Ryzen 5 and Ryzen 7 models are likely to be manufactured using a 3nm class node from TSMC, featuring a monolithic design. Enthusiasts will be particularly excited about the Ryzen 9 Medusa Point configurations, which are rumored to feature a staggering 22-core layout. This setup comprises a 12-core Zen 6 chip complex die (CCD), which is similar to the one used in the Venice architecture, along with four classic Zen 6 cores, four dense Zen 6c cores, and two LP cores. The multi-chip module (MCM) design is highly indicative, combining a 10-core chiplet with I/O and iGPU functions from the mainstream Medusa Point APUs and a robust 12-core CCD, likely fabricated using TSMC's N2 process. While specific details about the memory controllers, neural processing unit (NPU), and cache configurations remain undisclosed, it is clear that AMD is pushing the boundaries of what is possible in mobile and high-performance computing. The inclusion of dense and low-power cores alongside classic high-performance cores suggests a balanced approach aimed at optimizing both performance and power efficiency. However, competition in the APU market will be intense, with Intel's forthcoming Celestial architecture in Panther Lake looming on the horizon. Early specifications indicate that AMD's Medusa Point could outpace Panther Lake in CPU performance, although the exact timeline and competitive landscape will play a crucial role. If Medusa Point launches in early 2027, it could face off against Intel's Nova Lake, further intensifying the rivalry. To stay informed about the latest developments and insights into AMD's Medusa Point and other tech advancements, follow Tom's Hardware on Google News and make sure to hit the Follow button.

Related Links