The AMD Ryzen 7 5800X3D Review: 96 MB of L3 3D V-Cache Designed For Gamers
by Gavin Bonshor on June 30, 2022 8:00 AM EST- Posted in
- CPUs
- AMD
- DDR4
- AM4
- Ryzen
- V-Cache
- Ryzen 7 5800X3D
- Zen3
- 3D V-Cache
CPU Benchmark Performance: Power, Office, and Science
Our previous set of ‘office’ benchmarks has often been a mix of science and synthetics, so this time we wanted to keep our office section purely on real-world performance.
For the remainder of the testing in this review of the Ryzen 7 5800X3D, we are using DDR4 memory at the following settings:
- DDR4-3200
Power
Looking at the power draw between the Ryzen 7 5800X3D and the other chips tested, it is more power-efficient than the original Ryzen 7 5800X. This could be down to a load VID core voltage as the 5800X3D is actually clocked lower by 100 MHz at turbo clock speeds.
Office
In our office benchmark, the newer Ryzen 7 5800X3D performs similarly to the previous Ryzen 7 5800X processor.
Science
Our science-based benchmarks, for the most part, show that the Ryzen 7 5800X is slightly better computational-wise than the Ryzen 7 5800X3D. This is primarily due to the 5800X being clocked 100 MHz higher than the newer 5800X3D.
Where the extra L3 cache can benefit performance, it does, including in AI Benchmark, but overall the performance is very similar between both chips.
Facebook
Twitter
RSS
125 Comments
View All Comments
Stuka87 - Thursday, June 30, 2022 - link
Its not you, there is two full sets of them.Gavin Bonshor - Thursday, June 30, 2022 - link
Thank you, fixed 😊RBeen - Thursday, June 30, 2022 - link
Also does wonders for Planetside 2. I'm getting almost double the FPS from a 3600Xbrucethemoose - Friday, July 1, 2022 - link
Ah, the other game I wanted to see!Unsurprising, and it definitely needs all the CPU it can get.
bunkle - Saturday, July 2, 2022 - link
Nailed it. That's the biggest problem with all the games tested these days, they are single player games where CPU performance matters very little IMO.Games that are heavily CPU bottlenecked tend to be online FPS games with large player counts (~100) and lots entities to simulate such as Plantside 2 you mentioned and many Unreal Engine 4 titles: Hell Let Loose, Squad, Post Scriptum, Beyond the Wire, Holdfast Nations at War, Rising Storm 2: Vietnam, Squad, PUBG etc. to name but a few.
The problem is that it's very hard to reliably benchmark these games where it matters: in online gameplay without developer support. It's also where gaming has stagnated for the last 10-15 years. We have these amazing openworld maps that are completely barren with nothing happening in them and stuck with a limited number of players.
I realise that this isn't directly related to CPU performance and is also a software engineering challenge simulating world across multiple cores etc. but being able to showcase new CPUs and associated performance in these titles would probably help a lot with CPU marketing and would probably drive further innovation. Just look at UE5, theirs no mention of it's online capability or what new gameplay it enables, just more eye candy.
MadAd - Sunday, July 3, 2022 - link
Agreed, in an shooter like Planetside 2 the closer things like resist tables and default texture maps can be to the CPU the more fps can be gained on the graphic side. Having easily 200-300 players in a single fight over a base along with tanks, quads, troop transports and aircraft flying around you need the best single core processor you can get, which is why I switched to AMD with the 5 series and having more 3D cache just sweetens that pot.Slash3 - Thursday, June 30, 2022 - link
On page 5, it states memory used is DDR4-3200 CL40. I assume that's a typo, and that the usual JEDEC 3200 CL22 kit was used?Slash3 - Thursday, June 30, 2022 - link
Also, "insert analysis" at the bottom of page 7, and "DDR4-43200" at the top of page 8. ;)Glad to see the review up!
Gavin Bonshor - Thursday, June 30, 2022 - link
You are correct, was just a typoDevBuildPlay - Thursday, June 30, 2022 - link
Great article! I would love to see a similar article on the 3D cache differences on Epic under DB and other server workloads.