Ashes of the Singularity DX12 Benchmarks With AMD and NVIDIA

Jump To:

Ashes of the Singularity Benchmark II DX12 Testing

Stardock and Oxide Games released the Ashes of the Singularity Benchmark I in the Fall of 2015 and we ran some tests with the first non-synthetic Direcxt12 benchmark test and found some very interesting results.  Ashes of the Singularity uses the Nitrous Engine that is capable of using many CPU cores, so this benchmark should be a good way to look at how current video cards from AMD and NVIDIA will perform on a DX12 game title that has parallel rendering, asynchronous compute and even explicit multi-GPU support.  Last week we were given access to the new Ashes of the Singularity Benchmark II tool that has been updated to better replicate real world user scenarios.

Here’s what’s new:

  1. Explicit Multi-GPU. You can now insert an additional video card into your PC and increase performance by up to 2x. Explicit Multi-GPU allows gamers to use an AMD card and an Nvidia card in the same system.
  2. Significant general performance optimizations.
  3. The addition of the game’s Substrate faction in the benchmark.
  4. Increased the benchmark’s overall load to test expanded gameplay features.
  5. New graphics effects.
  6. Advanced use of D3D12 multi queue and signaling mechanisms. This is often referred to as asynchronous compute.

Ashes of the Singularity

We started to benchmark a handful of video cards on the new Ashes of the Singularity Benchmark II tool, but AMD was late to the driver game again and that Radeon Software Edition 16.2 video card drivers were optimized for this benchmark. Once we got done running the AMD benchmark numbers NVIDIA released GeForce 362.00 WHQL drivers, so we went back and updated all of our numbers again to ensure both AMD and NVIDIA were given a fair shake with the latest drivers. Since we were holding the article to ensure we were using the latest drivers from AMD what harm was there in delaying it a couple more days for NVIDIA.

DX12 GPUs Tested

The six video cards that we tested on DX12

For testing we pulled out six video cards that we wanted to look at with three being from AMD and two being from NVIDIA. We tried to pair cards together at price points to show DX12 performance at three different price points on current video cards.

Flagship Video Cards – $599+

High-End Graphics Cards – $300 – $350

Mainstream Graphics Cards – $189 – $229

Let’s take a look at the test system and then see how these cards performed at 1920 x 1080, 2560 x 1440 and 3840 x 2160 screen resolutions!

Jump To:
  • Matthew Curry

    well looking like my 380x wasnt such a bad buy lol. crossfiring it later this month and got ashes of the singularity in a free offer with newegg. cant wait to see the benchmark.

  • Hapi hap

    cant wait to see other games with this new flavor of multi-gpu on dx12, has there been any other titles announced?

  • NMRH

    Well, if anything, the 390 its performing very nicely for its price.

  • Sean Easterling

    But how is the mult-gpu scaling compared to SLI or Crossfire? Could you do 2x FuryX vs 2x 980ti multi-gpu benches?

  • shadowhedgehogz

    Considering Nvidia hasn’t enabled Async (if they even can..) the 980 Ti is doing really well, close to the Fury x.. Great news for owners of that card, though the 970 is looking pretty weak and likely the 980 won’t be much better.

    I guess the 980 Ti just has enough brute force strength to put in a good showing.

    • Diego Paolini

      they using 980ti amp extreme that is one of best 980ti factory overcloket… with reference gap its about 20-30%

      • Robdarian

        At close to the same price, no one is buying reference cards. Same as when everyone bitched about the 290x reference overheating, no one was even buying that card.

  • Deregtz

    What about integrated graphics combined with AMD or Nvidia? Iris IGP is pretty powerfull these days.

    If you test that it would be pretty interesting.
    Now its just another combine a AMD and Nvidia GPU article..

    While most people have want to know how integrated graphics will work together with a GPU.

    • Daniel Anderson

      I’m honestly curious about Intels iGPU and DX12 w/ and w/o dGPU’s.

      • Jeffrey Byers

        It’s been tested and it DID benefit. It added over 60% of it’s processing power. The better the dGPU the less noticeable the improvement is but it definitely can benefit.

        For example, if the iGPU alone got 10FPS and the dGPU alone got 50FPS then together it was 56FPS.

    • Domaldel

      From what I’ve heard it’s a bad idea to combine a low performance GPU with a high performance one in this benchmark.
      You end up with the dGPU performing worse then alone.
      However some of the AMD APUs apparently manages to improve the performance with a low/mid end dGPU (with dx12)

      (There was an article about that stuff somewhere)

  • Coach

    Love the article. It is very interesting to see the results, especially the combo of the two brands.