Benchmarking Showdown & DiRT Rally

Showdown

Showdown is not a game, but rather a 2014 Unreal Engine 4 demo by Epic that was the genesis for BulletTrain and finally Robo Recall, one of this editor’s favorite VR games.  This demo was incredibly popular at trade shows and is still used for demonstrating what VR can do as there are a lot of effects and interactivity.   However, this demo is no longer considered demanding as some of the particle effects appear to be 2D sprites instead of actual 3D geometries, and there are no shadow casting lights (i.e. lighting is “faked”) even though the anti-aliasing is at 4xMSAA.

Although Showdown may be trivial for even a GPU like GTX 1060 or an RX 580 now, it will still show differences in the amount of performance headroom that each GPU has. Here is the professionally captured demo of Showdown in VR as uploaded by Epic.

Of course, our own video capture is taken from the left lens of the HMD and it will not be as clear as Epic’s capture as the lenses provide their own distortion and it also cuts down the viewing area, and in the interest of repeatable benching, we look straight ahead during our benchmarked Showdown runs.

Here is our video Showdown benchmark run by the GTX 1080 Ti.

As you can see there is a lot of performance headroom.   Now compare with the FCAT VR frametime graph.

No frames were dropped.  203.3 unconstrained FPS were averaged by the GTX 1080 TI.

Here is our video Showdown benchmark run by the GTX 1080.

As you can see there is a lot of performance headroom also with a GTX 1080.   Now compare with the FCAT VR frametime graph.

No frames were dropped as the GTX 1080 averaged 182.24 unconstrained FPS.

Here is our video Showdown benchmark run by the GTX 1070.

As you can see there is still decent performance headroom.   Now compare with the FCAT VR frametime graph.

No frames were dropped and the GTX 1070 managed 159.2 unconstrained FPS.

Here is our video Showdown benchmark run by the EVGA GTX 1060 SC.

As you can see there is a lot of performance headroom.   Now compare with the FCAT VR frametime graph.

Although two frames were dropped, the GTX 1060 can manage 128.5 unconstrained fps

Here is our video Showdown benchmark run by the FURY X.

As you can see there is still some performance headroom.   Now compare with the FCAT VR frametime graph.

Although 3 frames were dropped by Fury X, it manages 124.1 unconstrained FPS 

Here is our video Showdown benchmark run by the RX 480 at RX 580 clocks.

As you can see there is a little performance headroom.   Now compare with the FCAT VR frametime graph.

Although 5 frames were dropped, the RX 480@580 clocks averages 105.8 unconstrained FPS

Here is the FCAT VR charts of all 6 cards but without the interval plot bars.  There is very little reason to show them as few frames were dropped.

As you can see from the videos and the FCAT VR graphs, Showdown is not demanding, yet we can clearly see the ranking of our video cards from top to bottom – or actually bottom to top, as lower latency is ideal.  Make sure to read the chart noting that the frametimes that are lower on the chart are delivering frames more quickly than those higher up the chart.

Watching the Motion to Photon Latency as well as the Performance Headroom in the Overlay, we can accurately gauge performance.  Every time the App Frame Rate (Hz) green line dips, the framerate drops to 45 FPS; and when it is a solid yellow line, and ASW is delivering a reprojected frame to maintain the illusion of 90 FPS and it works well to keep one from getting VR sick.

DiRT: Rally

DiRT Rally is special as it has a built-in benchmark that is 100% repeatable.  This benchmark is quite accurate and it will relate to the FCAT VR graphs.  Whenever a framerate drops below 90 FPS, the Test Summary of the built-in benchmark accurately shows the minimum as 45 FPS (with ASW reprojection).  Unfortunately, it  does not display unconstrained FPS. It’s also a very fun game for either PC gaming or for VR as a very demanding game that requires the player to really learn the challenging road conditions.  Best of all, DiRT: Rally has a lot of settings that can be customized although we benchmarked only the “High” and the “Ultra” presets

Here is the DiRT: Rally benchmark run by the GTX 1080 Ti at Ultra settings.

As you can see that although there is a lot of performance headroom with 117.7 unconstrained FPS, some frames were dropped in this 3-1/2 minute in-game benchmark. 

Here is our video DiRT: Rally benchmark run by the GTX 1080 at Ultra settings.

The GTX 1080 managed an average unconstrained fps 93.2 which mostly kept it out of reprojection.

Here is the DiRT: Rally benchmark run by the GTX 1070 at Ultra settings.

Ultra settings with GTX 1070 produced 76.7 unconstrained FPS which means a lot of ASW, but on the High preset it managed 97 unconstrained FPS with very few dropped frames.

Here is our DiRT: Rally benchmark run by the EVGA GTX 1060 SC on the High preset.

The GTX 1060 is a little too weak for High and we would recommend dropping some settings.

Here is our DiRT: Rally benchmark run by the FURY X.

The Fury X managed 84.6 unconstrained framerates which allowed it to play at High.

Here is our video DiRT: Rally benchmark run by the RX 480 at RX 580 clocks.

The RX 480@580 clocks only manages 71.8 unconstrained fps with a ton of ASW generated synthetic frames and a few dropped frames besides.  Lowering settings is the best way to play.

Here is the FCAT VR chart of 3 cards with the intervals plotted using the Ultra preset.

Now here is the High preset  with 4 cards – we are repeating the GTX 1070 on High settings.

Only the GTX 1070 and the Fury X can handle high settings in the DiRT Rally benchmarks; lower setting to play with a GTX 1060 or with a RX 580.

Let’s look at Alice VR and Chronos next.

Oculus Rift

9

Overall Score

9.0/10

Pros

  • Costs $600, $200 less than the Vive
  • The Touch controllers are awesome
  • The HMD is lighter than the Vive
  • It is easy to play seated

Cons

  • $600 is expensive
  • You need a powerful PC
  • Roomscale VR is not ideal

2 COMMENTS

  1. AMD doesn’t have a answer to the 1070 or 1080 yet. Go ahead and benchmark the $200 cards though. I wouldn’t want em for VR. Also, It’s completely stupid to compare a 580 to a 1080.

Comments are closed.