AMD’s Zen 5-based Ryzen 9000 processors will land next month, offering an average 16% IPC gain on the TSMC 4nm (N4P) node. The chipmaker’s official benchmarks claim up to 23% lead for the Ryzen 9 9950X over the Intel Core i9-14900K in gaming workloads. The content creation performance is even more impressive, beating the 14900K by up to 55% in Blender and Handbrake. But when asked about the gaming capabilities of these chips versus the Ryzen 7 7800X3D, AMD’s response was this:
It’s (Ryzen 9 9950X) faster than the competition in our tests. X3D is still the king of the hill, but by a much smaller margin than typically between X3D and non-X3D. So a 7800X3D would, yes, be faster than 9700X, but maybe not by as much as you would expect.
And then when it comes to X3D, and I’ll just get around that now, we’re super committed to X3D. In fact, we have some really, really cool updates to X3D coming. So we’re working on iterating and not just rehashing it
Donny Woligroski, Senior Technical Marketing Manager, Ryzen (via Tomshardware)
The Ryzen 7 7800X3D will retain a slim lead over the Ryzen 7 9700X and the 9950X in gaming, but that won’t last long. The Ryzen 9000X3D CPUs are expected to land at CES 2025, perhaps briefly receding the crown to Intel’s Arrow Lake-S processors.
AMD promises some “really, really cool updates” to the upcoming Zen 5 X3D chips, but we’re currently clueless. The Ryzen 7000X3D processors featured a higher stacked L3 bandwidth of 2.5 TB/s, up from 2 TB/s on the Ryzen 7 5800X3D. TSMC’s 9-micron TSV connections were used on both generations, and this time we might see an upgrade on that front, boosting data throughput and latency.
The Ryzen 9 7900X3D and the 7950X3D feature a vanilla and a 3D stacked CCD. The former is more or less useless in gaming but comes in handy in content creation. The reason? Keeping all your game code in one unified cache pool reduces latency and pipeline stalls. Two pools would be less effective, but the Zen 4 X3Ds feature one large L3 pool on one die and a regular L3 pool on the other.
This (at times) divides the data required by the same thread into different cache reserves, thereby increasing latency. Programming games to limit thread data to the same CCD helps, but there are limits to this method. There’s also the matter of thermal dissipation.
Stacking an L3 cache die on top of the 4nm CCD worsens thermals. Consequently, the maximum temperature of the X3D chips is lower than the regular Ryzens, and overclocking is blocked off. It’ll be interesting to see which path AMD takes with the Ryzen 9000X3D lineup. Our guess is an upgrade to the TSVs, and potentially larger L3 cache dies. Multiple cache dies won’t be as effective.
