It has been almost 10 years since the launch of AMD's Ryzen processor, and the CCD architecture is finally at the stage of major changes. Zen6 with 2nm process will be launched this year, and this time there will be a 12-core CCD computing core. AMD officially announced the Zen6 architecture at the CES show earlier this year, but it was the EPYC product line Venice processor targeted at the server/AI market. This generation will be designed to focus on high performance and throughput, with a 30% increase in thread density and a 70% increase in performance and energy efficiency.
The server version of Zen6 will have up to 256 cores this time, and the desktop version of Zen6 will be codenamed Olympic Range, and its architecture will also be significantly changed.Netizen HXL broke the news that its CCD core will be upgraded to an 8-core architecture, equipped with a 48MB L3 cache, which is 50% higher than the current Zen5.
More importantly, the core area only increased by 5mmm2 in the 2nm process.Increased from 71mm of the previous generation to 76mm2 of Zen6,The area efficiency is very high, after all, the process has been upgraded for almost 2 generations.
AMD's Zen architecture has been around for nearly ten years. This is the second major change in CCD scale, and it is also the largest scale improvement. Mainstream desktop platforms can easily achieve 24 cores and 96MB L3 cache using 2CCD configuration, and higher-end ones can achieve 48 cores and 192MB L3 cache.
If it is the X3D series, HXL mentioned that Zen6 CCD also has a complete 3D V-Cache cache function. Each CCD can be equipped with up to 144MB L3 cache, and 2CCD can achieve 288MB L3 cache.
Considering that Intel will release it at the end of this year and go on the market next year,Nova Lake-S desktop version is also rumored to have up to 288MB cache and up to 52 cores.AMD and Intel are bound to have a battle over enthusiast-level gaming CPUs.