AMD Ryzen 9 9950X3D2 With More 3D V-Cache May Be Coming Plus Another X3D CPU

There was a time when processor companies released new CPUs a full generation at a time. In the last several years though, AMD has thrown that practice out the window. New CPUs come from the company whenever they're good and ready, it seems, and supposedly, there are two more Zen 5 desktop chips with 3D V-Cache on the way: the Ryzen 7 9850X3D, and the Ryzen 9 9950X3D2.

This information comes from occasional leaker chi11eddog (@g01d3nm4ng0 on Xwitter), who simply says "New Ryzen coming..." before delivering the deets on the supposed desktop CPUs. The Ryzen 7 9850X3D is reportedly the exact same thing as a Ryzen 7 9800X3D, except with a 400MHz bonus to its boost clock, raising the peak speed for the new chip to 5.6 GHz. That's higher even than the Ryzen 7 9700X, which tops out at 5.5 GHz.


The really interesting chip is the purported Ryzen 9 9950X3D2, though. This chip deviates from extant Zen 5 processors in a few ways: a 200W TDP, which is novel, and 192MB of L3 cache. That's right: this is the mythical "dual V-Cache" CPU, with two 8-core Zen 5 CCDs each resting atop a 64MB SRAM chiplet. Even with the TDP bump, the peak boost clock supposedly still takes a 100MHz chop versus the Ryzen 9 9950X and Ryzen 9 9950X3D, which actually have the same 125W TDP and 5.7 GHz boost.

It's really not clear that an "X3D2" processor will have significant advantages over the standard X3D chips. Certainly, at least, for gaming, the Ryzen 7 9800X3D with one CCD and one V-Cache chiplet offers essentially the same or better performance versus the Ryzen 9 9950X3D. However, there are a scant few non-gaming workloads that specifically benefit from 3D V-Cache, including code compilation, database analysis, computational fluid dynamics, and finite element analysis. This chip could end up being a killer workstation CPU.

As is usually the case, chi11eddog provides no source or proof for his claims, but this is not the first time he's promised these exact processors. Back in August, he claimed that these chips were on the way, but a known leaker on the Chinese-language ChipHell forums, wjm47196, retorted that chi11eddog's information was "fake". Now that he's back with the same story, should we take him at his word? Doubling down this way will have dire consequences for his credibility if the chips don't materialize sooner or later.