DirectX Comparison Tessellation

Testing Methodology

HAWX 2 plays using either DirectX 9 or DirectX 11. By default the game launches in DirectX 9. To play using DirectX 11 you must launch the game using a different executable in the install directory "HAWX2_DX11.exe". HAWX 2 includes a benchmark routine that can be launched from inside the game or as a stand-alone application (also requiring a special shortcut). This is a decent benchmarking tool if you want to evaluate system improvements or to tune your graphics settings to an enjoyable level. But outside of this there are some flaws that hinder its ability to compare across systems.

First, the HAWX 2 benchmark simply isn't rigorous enough to put a demand on newer video cards. At a 2560x1600 resolution, 8x AA, 16x AF, and tessellation "on" a GTX480 nets 78 FPS while the HD5870 nets 41FPS. In other words, even the most strenuous game settings can easily be rendered by newer video cards. Likewise even mobile GPU's can get decent frame-rates at commonly used graphics settings (see ASRock HTPC).

The second issue is the imbalance between NVIDIA and ATI cards. Most of the benchmarks I've seen between the HD5870 and GTX480 were much closer than the near 50% gap listed above. The worst I had seen on any previous benchmark put the HD5870 falling 30% short of the GTC 480. But you probably could have expected this discrepancy in both the game and the benchmark since the NVIDIA logo is plastered all over the game and ATI is not to be seen.

From the perspective of a game developer who needs to cut costs and meet arbitrary release deadlines it probably makes sense to rely more heavily on the company that consistently leads the GPU market in performance. But for a system tester trying to give a balanced view of different hardware components a biased testing tool just can't be used.

In the end the game was tested with different cards and different graphics settings to evaluate the impact on the gaming experience. Screenshots from each setting were taken and displayed for you to make up your own mind on the difference offered.

Test System

• Motherboard: Gigabyte GA-MA790X-UD4P
• Processor: AMD Phenom II X4 965 BE CPU HDZ965FBGIBOX
• System Memory: OCZ Gold 4GB 800MHz DDR2
• Disk Drive: SAMSUNG HD251HJ (232 GB)
• PSU: Antec EA500 EarthWatts 500W PSU
• Operating System: Windows 7 Professional
• Video Cards:
  
  • PowerColor Go! Green Radeon HD5670
  • Palit GeForce GTX 260 Sonic 216SP

Tessellation Results

The first test was made using an HD5670 and full graphics settings. One screenshot was taken with tessellation on and another taken with tessellation turned off. The DX11 benchmark was also run at both settings to see the impact on frame-rate. With tessellation on the HD5670 drew 55 FPS max and 30 FPS average. With tessellation off it drew 91 FPS max and 51 FPS average. The images are displayed below.

I left the images unlabeled on purpose. Without reading ahead can you tell which picture is which?HAWX 2 was written to take advantage of DX11 tessellation.Terrain is formed by a rough height map spliced together with a bi-cubic filtering then jittered using fractal noise. After all this the tessellation is applied to draw 1.5 million triangles per frame, all on the GPU. The result with or without tessellation is quite impressive. This is definitely the best terrain I've seen in a flight game. I honestly find it difficult to see much difference between the tessellated scene and the non-tessellated scene.

Now for the answer; the image on bottom is the tessellated landscape. The first thing I notice is the difference in lighting. The non-tessellated image looks like the landscape is dark and blurry, like the features are smeared together. The tessellated landscape has better distinctions between small peaks, valleys, cracks and such. It makes things look more alive. But I hardly notice any of this while I'm looping endless circles. Next we'll show some larger images in which the other graphics settings are also varied.

<< Previous Page - Next Page >>