Performance and power consumption
The first thing to notice about the 9850X3D is that its multi-core performance is essentially indistinguishable from the 9800X3D. If anything, the 9800X3D seems marginally better behaved in our Handbrake video encoding tests, which could come down to anything or nothing—maybe it’s an actual difference between the chips, maybe it’s the silicon lottery, maybe it’s something else. Overall, it’s mostly a wash.
The 9850X3D does improve significantly on the 9800X3D’s single-core performance, closing the gap with both the Ryzen 9900X3D and 9950X3D (both of which have regular Zen 5 cores with no 3D V-Cache) and the 9700X (which has no 3D V-Cache at all).
There’s a world in which AMD can get this extra performance for “free” without changing anything about the architecture or manufacturing process. AMD could have been “binning” silicon lottery winners, or the reliability of the manufacturing process could have improved enough to allow AMD to hit better numbers that weren’t as consistently achievable a year ago.
But it looks like AMD improved the 9850X3D’s single-core performance mainly by making it behave more like a non-X3D chip. The chip’s power consumption while gaming suggests this is more or less what’s happening—the 9850X3D’s CPU package power while gaming is some 25 or 30 W higher than the 9800X3D playing the same games, despite the single-digit-at-best performance improvement (in both our testing and AMD’s advertising).
While the 9850X3D’s power consumption while gaming isn’t wildly out of step with the 9700X’s or the 9950X3D’s, it does make the tiny performance gains on display here hard to get excited about.
And this is on top of the cost trade-off you’re already making when you buy an X3D-series chip. Game performance is always impressive, but you notice those benefits the most in situations where your CPU, not the GPU, is the performance bottleneck.