Gaming Performance Continued
We are testing only at 1080p, to isolate CPU performance. We are using an NVIDIA GeForce RTX 4090 Founders Edition video card with driver version 565.90. All games are being run in their “Ultra” game settings, at native 1080p resolution, and with no ray tracing or upscaling. In some games, we have set the desktop also to 1080p for Windowed/Borderless Window games. Importantly, we are using ALL manual run-throughs for every game here, not a single built-in benchmark was used, these are all manual run-throughs actually in the game, playing the game.
All CPUs are tested in Windows 11 power profile of “Balanced.” Where noted, Intel Arrow Lake CPUs are shown in both Balanced (B) and Best Performance (BP) power profile modes. As you look at the graphs, pay special attention to the AMD Ryzen 7 9800X3D (blue bar) versus AMD Ryzen 7 7800X3D (light gray bar), to see the generational uplift.
Cyberpunk 2077
In Cyberpunk 2077 the new AMD Ryzen 7 9800X3D is 9% faster than the Ryzen 7 7800X3D and 11% faster than the Ryzen 9 7950X3D. It is also a pretty jaw-dropping 33% faster than the Ryzen 7 9700X for gaming here. Compared to the competition, it demolishes the Intel Core Ultra 9 285K by 44% at its best, and 35% faster than the Intel Core Ultra 9 i9-14900K. This graph just really paints the whole picture very clearly overall.
Dying Light 2
Dying Light 2 is another game that does quite well for showing CPU performance differences. The new AMD Ryzen 7 9800X3D is 15% faster than the previous generation Ryzen 7 7800X3D and 7950X3D. It’s also a pretty big 45% performance improvement compared to the Ryzen 7 9700X in this game. You can really see how the X3D chips boost performance in this game, those without do fall much farther behind. The AMD Ryzen 7 9800X3D is 47% faster than the Intel Core Ultra 9 285K at its best, or 54% faster at the same power profile plans. The AMD Ryzen 7 9800X3D is also 45% faster than the Intel Core i9-14900K.
Hogwarts Legacy
Hogwarts Legacy is also a game that can be greatly affected by CPU performance, especially in Hogsmead, where we test. The X3D parts have a leg up in performance for sure in this game. The new AMD Ryzen 7 9800X3D is a very impressive 25% faster than the previous generation Ryzen 7 7800X3D and 7950X3D. It’s also a very massive 53% faster than the Ryzen 7 9700X in gaming performance. This puts it ahead of the Intel Core Ultra 9 285K by 53% at its best power profile, or a melting 70% faster at the same power profile plan. The AMD Ryzen 7 9800X3D is 34% faster than the Intel Core i9-14900K.
Starfield
Starfield is also affected a lot by CPU performance, and X3D chips. The new AMD Ryzen 7 9800X3D is 14% faster than the previous generation Ryzen 7 7800X3D and 7950X3D. The AMD Ryzen 7 9800X3D is also 34% faster than the Ryzen 7 9700X, which is just impressive. Compared to Intel, the AMD Ryzen 7 9800X3D is 19% faster than the Intel Core Ultra 9 285K and 14900K.
Discussion (19 replies)
Join Discussion →Great review and the 9800X3D looks like a beast.
Yep, keeping my eye on this one and can't wait for the 9900X3D/9950X3D reviews to make my final decision for my 1st AM5 build.
After this review I am more excited about this gen than I have been before. The extra die space from these X3d's is having some add on benefit for performance that is hard to quantify (more space for heat?) but I don't care. This is kind of awaesome.
Just curious, what are you referencing here?
For pure gaming I'd still go with the 7800X3d, but the 9800X3D is better than the 9700X in every way. I don't really think we'll ever see a 9800X.
Hopefully we'll eventually see a 9700X3d or even a 7700X3d that's I would get myself.
I wonder what OC will bring to the table.
If going by recent rumors, anything from 200-400 MHz but so far as to how that will translate, added power draw, heat, etc., haven't heard anything credible yet.
I think he means this from Brent's review.
"In the 2nd Generation 3D V-Cache, AMD has re-designed the cache so that it is the same size as the CCD itself, eliminating the need for structural silicon."
@Brent_Justice
"In subsequent follow-ups, we will deep dive into gaming performance a lot more, with a very gaming-focused review with a lot more games, so stay tuned." - I bet you'll have a lot of fun doing this, very jealous!
Sorry if I missed mention of it, is there a separate review for that Trident mem kit coming up? I was doing some research on it this weekend as I prepare for the new build.
AMD is still using the same IOD - so 6000 C30 is basically it without tempting fate, same as at Zen 4 launch.
Some folks in the OC communities have been working on faster memory speeds but I haven't seen any real evidence of gains in effective performance.
Don't have one planned, but I've been using a similar kit since Zen 4 launched, it's my standard RAM kit for the AM5 platform, and I've been using it for every GPU review. Solid kit, no complaints, works perfectly great, it's the right configuration for the best Ryzen performance.
To be honest, I am quite looking forward to a 20 game performance roundup!
Done it before: https://www.thefpsreview.com/2023/11/20/intel-core-i9-14900k-vs-amd-ryzen-7-7800x3d-gaming-performance-in-20-games/
Basically in the review it was noted that the 3dvcache space is the same as the die space. Meaning that there is heat dissipation that can come downward into the 3dvcache layer. Allowing the actual processors to be pushed further.
nice work guys!
It would appear that placing the 3D cache on the bottom is allowing it to boost closer to it's full potential. This is good news. Now I can only hope even more than the 9950X has dual 3D cache on both of it's CCDs.
That makes a little bit of sense; still, it isn't likely to have near as much of an effect as having just gotten the 3D V-cache out of the way of the heatspreader IMO.
Disappointed in the lack of 2560x1440 and 3840x2160 numbers. I've done a lot of this type of testing over the years and 1080P just doesn't tell the whole story. Conventional wisdom says that when your GPU bound the CPU doesn't really matter, but I found that wasn't always the case. Often you'd get averages that seemed pretty close but the lows were lower and the highs higher on some CPU's.
On games like Cyberpunk 2077 where a few frames can make the difference between a playable frame rates and not, (even on a 4090) the CPU might make a difference. It sure did with earlier Ryzen family CPU's I tested.