These benchmarks do not represent real-world gameplay by any stretch of the imagination. These tests are conducted using built-in benchmarking tools utilizing the game engine. These tests are designed to stress the CPU’s power system as well as ensure proper functionality. These are all run at CPU limited resolutions to try our best to remove the video card as a bottleneck.
One additional note about gaming performance is that these are average values. They do not tell the whole story. Specifically, Destiny 2’s low’s and high’s help tell a very different story than what the average reports are. Fortunately, while the lows are considerably lower than that of Intel CPU’s, the amount of frames that drop into the lowest range are extremely few.
As stated in previous articles, we are transitioning to providing more comprehensive data for gaming benchmarks. Leveraging some of the built-in tools, we are showing minimums, averages, and maximum frame rates. Frametimes are something we will work on including in the future as well. Unfortunately, we didn’t have that much lead time from Intel on this review.
All of our test systems score fairly well here. However, at stock speeds, the Ryzen 9 3950X falls behind by a small amount. I can only surmise that this is due to the base clocks it uses, or that for some reason it doesn’t boost all that well in this application. Manual overclocking resolves this issue and the score is only 100 points or thereabouts off of the 9900K, which is the gaming champion. Of course, 3D Mark results do not always line up with expectations and once again, the 10980XE does very well here and achieves the highest score when overclocked. At stock speeds it is just behind the Ryzen 9 3900X.
In our Heaven benchmark, we see rough parity with most of our test systems. This is one area where our stock 10980XE gets beaten pretty badly. It’s also worth noting that I retested the 9900K with the same GPU as the rest of these systems. Previous results used a 2080 Ti and not a 2080 Super. As a result, it takes a pretty big hit. Why it comes up short, I’m not quite sure. However, the 3900X proves to be the sort of high-end sweet spot for a gaming machine. The 3950X does well, just not as well as the 3900X does.
This test often yields interesting results on multi-core and HEDT CPU’s in particular. We see vastly superior results from the Core i9 9900K and even the 10980XE when it is sufficiently overclocked. In fact, its a rare case where the 10980XE ends up being faster than the 9900K. The Ryzen 9 systems all perform within a margin of error and are functionally identical here. Clock speed differences don’t seem to matter all that much here and that’s likely due to the boosting behavior of Ryzen CPU’s, in general, being inconsistent, or bursting to the highest frequency for short durations and then dropping back to around 4.0GHz.
I had some trouble with our 9900K test rig as the 2080 Super was overheating. As a result, I couldn’t get Hitman 2 to run reliably here. This is something I’ll have to deal with later, so these results have been omitted. As you can see, the 10980XE is by far faster in this test than anything AMD has to offer. Overclocked, the 10980XE is a monster. Weirdly, the 3950X once again falls behind the 3900X. I didn’t expect it to lose so much ground to the 3900X, but there it is.
I apologize for the slightly weird scaling here. I did this to fit all the results on the chart where you could see them. The 10980XE’s maximum FPS results are quite literally in the stratosphere. This is common on HEDT platforms with Destiny 2 as this is something the Threadripper 2920X did in the past. As you can see, the AMD systems produce very similar results. The 3950X has a slightly higher minimum when overclocked, but loses a little in the averages and maximum frame rates. Again, the 10980XE proved to be the fastest solution.
What’s even stranger is the fact that the Ryzen 9 3900X, when retested shed its poor minimum FPS scores and achieved a solid 64FPS minimum. This is incredibly different than the data I got using a faster video card back when a workaround had to be employed to make this game even run on that CPU. Out of these CPU’s, it is the 3900X that would provide the best experience in my opinion. Keep in mind that the test system for the 3900X and 3950X is the same. We only swapped CPU’s and nothing else.