Computer Base tests DX12’s mettle in The Division

Microsoft's DirectX 12 API gets more popular every day. The latest AAA title to add support for Redmond's low-level graphics API is Ubisoft's The Division. German hardware site Computer Base took some time over the weekend to run some performance analysis comparing different combinations of DX11 or DX12, Intel Core i7-6700K and AMD FX-8370 CPUs, and four different graphics cards from Nvidia and AMD. Roughly put, the results show that performance in The Division with DX12 is improved with all hardware combinations except for the Intel Core i7-6700K with Nvidia's GeForce GTX 1060.

The biggest gain was the 13% increase shown by the Intel CPU with the GTX 1080. The AMD Fury X and RX 480 enjoyed improvements of four to nine percent regardless of CPU choice. The biggest changes in performance were at the 1920 x 1080 resolution, with deltas shrinking as the resolution went up. Some of the measured performance boosts were within the margin of error, though, like as the one-percent increase shown for the FX-8370 with the GTX 1060.

Computer Base has more graphs, including some frametime charts, as well as additional commentary on their findings, auf Deutsch, of course. We are living in the 21st century and online translation tools do a pretty good job these days, so go take a look.

Comments closed
    • Topinio
    • 3 years ago

    Saw this last week, what’s most interesting to me is that the FX-8370 seems to be (finally?) fully competitive with Intel’s best (and much newer) desktop CPU in what should be a non-GPU bound gaming benchmark.

    Particularly with the AMD GPU’s, and really with everything but the GTX 1080, there is nothing to gain from the 2 years newer Intel chip. Mad.

    [url<]https://www.computerbase.de/2016-12/the-division-dx12-benchmark/#diagramm-the-division-mit-directx-12-1920-1080[/url<]

      • chuckula
      • 3 years ago

      Really? That chart shows a 13 FPS (actually almost 14 FPS) lead for the 6700K under DX12 at a 1080p resolutions and over 8 FPS even in DX11 (note the [i<]bigger[/i<] lead in the more modern DX12 that is theoretically supposed to favor the 'moar coars' AMD architecture... at least in some theories). That's a pretty major difference in a video game where the GPU normally reigns supreme in being the limiting factor. Additionally, at 100FPS, the 6700K may even be running into GPU limited territory even for the GTX-1080.

        • Topinio
        • 3 years ago

        It showed leads for the 2 years newer Core i7 of only 1.5, 0.6, -0.7, -0.8, 2.7, and 1.4 FPS for the other 3 GPUs.

        It’s clearly not GPU-limited at 48-49 FPS on the weakest card (GTX 1060).

          • chuckula
          • 3 years ago

          [quote<]It's clearly not GPU-limited at 48-49 FPS on the weakest card (GTX 1060).[/quote<] You got any evidence for that statement? You do realize that in your attempt to hurl insults at the 6700K you are basically saying -- without evidence -- that the GTX-1060 ought to be performing better than it is against the Rx-480. Are you [b<]REALLY[/b<] sure you want to make that statement about Nvidia GPUs just to reinforce your point that Piledriver is really awesome? I mean, it's OK, you can let it go, even AMD has thrown Piledriver under the bus. You're allowed to copy-n-paste the new propaganda that every Zen chip is faster than the 6900K, go right ahead. From what I see, there's one GPU here, the GTX-1080, that isn't being bottlenecked at these resolutions and the 6700K is fully capable of taking advantage of it.

            • Redocbew
            • 3 years ago

            You are both reading what you want to read into these numbers. Multithreaded performance was never a problem for the bulldozer derivatives, so it makes perfect sense that a more thoroughly multithreaded game is going to perform better there. It’s also no surprise that DX12 performs well on the current batch of Radeons. We’ve known for a while that it would.

            With that said, “better” is relative, and it’s not like multithreaded performance in general is something AMD can make up a silly name for and try to make it their own thing. Furthermore, a complete migration to DX12 is still a long way off, so anyone who plans of keeping a piledriver chip around in hopes it’ll age like wine because of DX12 is in for a disappointment.

      • Waco
      • 3 years ago

      Keep in mind average framerates do not tell the whole story…

    • AnotherReader
    • 3 years ago

    However, at higher resolutions, AMD’s cards gain some performance though it could be within the margin of error. Using the best APIs for each card, at 1440p, Fury X has 83% of a GTX 1080’s performance if one goes by fps. At 1080p, the RX480 is 8% faster than the 1060.

    • AnotherReader
    • 3 years ago

    It would be nice to see an explanation of why the DX12 version brings greater gains to the 1080 than the Fury X. In August, [url=http://gpuopen.com/tag/warhammerwed/<]there was a series of posts by Tamas Rabel, Lead Graphics Programmer of Total War: WarHammer at GPUOpen[/url<] that explained how the engine changed in DX12.

    • tipoo
    • 3 years ago

    Biggest benefit to Nvidia. Iinteresting. Does this use async as well? I don’t really know how to ask in the ‘Kommentare’ 😛

    • chuckula
    • 3 years ago

    [quote<]The biggest gain was the 13% increase shown by the Intel CPU with the GTX 1080. The AMD Fury X and RX 480 enjoyed improvements of four to nine percent regardless of CPU choice. [/quote<] [url=https://www.youtube.com/watch?v=OTWZltMeorY<]Cleveland is not amused[/url<]

Pin It on Pinterest

Share This