Scope, Test Setup & Methodology
To help keep the article at a manageable scope, we are only going to explore the current state of DX10 from the perspective of the end consumer. Anything that is or should be transparent to the consumer, such as API optimizations that let developers program more efficiently, will not be covered. Note that this is not necessarily the same as exploring the current end user experience. We want to focus on examining the potential of currently available DirectX 10 hardware and software. We will focus our attention on two fronts; performance and image quality. Specifically, we are interested in examining the differences in performance between DirectX 9.0c and DirectX 10 and the image quality enhancements of DirectX 10, if any.
Furthermore, we will only explore the performance and image quality of currently available games on currently available hardware. We chose five video cards and five games that we believe will best represent the currently available selection of DirectX 10 hardware and software for use in our testing. However, we decided not to include any entry level DX10 hardware, because today's offerings are not well suited to cutting-edge gaming.
ATI vs NVIDIA: While the purpose of this article was to look at DirectX performance, some people will no doubt try to draw conclusions about which brand is better from our data so we might as well address the issue. First and foremost, we want to point out again that we made no effort to create a fair comparison between ATI and NVIDIA in our tests. The test setup for each game was chosen to try and illustrate how current generation graphics hardware will handle upcoming DX10 titles and the five video cards in our test were chosen primarily because we thought they represented the most relevant price points at the time the article was conceived.
Core 2 Duo E6850
GeForce 8800 GTX 768MB
GeForce 8800 GTS 640MB
GeForce 8600 GTS 256MB
Radeon 2900 XT 512MB
Radeon 2600 XT 256MB
Call of Juarez
Company of Heroes
Lost Planet: Extreme Condition
World in Conflict Demo
However, apples-to-apples tests are not without merit. As previously noted, one of the promises of DX10 is improved performance and efficiency compared to DX9 in certain situations. These efficiencies could result in better overall performance in an apples-to-apples situation where DX10 rendering proves to be faster than DX9 rendering. While we did not perform a full set of apples-to-apples benchmarks for every game, we kept an eye out for games that showed performance benefits under DX10 during our testing and performed additional apples-to-apples comparisons when necessary.
While our focus is on the current state of DirectX 10, we wanted to keep an eye on the future and how well the current batch of DX10 capable hardware will be able to handle upcoming titles. After all, many of the most hotly anticipated DX10 capable titles have yet to be released. In order to attempt to simulate the level of strain that future titles may put on hardware, we performed our tests with each game set to a high level of image quality with anti-aliasing and anisotropic filtering enabled wherever possible. Specific video settings and other test information for each game can be found on their respective pages.
A Cautionary Note Regarding Performance Test Results: Generally, we used the highest settings possible for each game, even if cranking a specific setting doesn't necessarily result in an appreciable image quality improvement. Usage of high video settings meant that our test system was put under a considerable amount of strain and performance numbers are expected to be on the low side. We designed our tests to illustrate the competency of currently available DX10 hardware and games and the test results may not necessarily represent a typical end user experience.
Image Quality Tests: For each of the five games in our test, we compared the image quality of DX9 rendering and DX10 rendering and attempted to find as many differences as possible. All image quality tests were performed with a GeForce 8800 GTX video card and the same test system configuration used in the performance tests. The video settings in each game were cranked up to their maximum levels and anti-aliasing as well as anisotropic filtering were enabled wherever possible.
Screenshots of each game using both DX9 and DX10 rendering were taken with Fraps at a resolution of 1920x1200. The screenshots were saved by Fraps in bmp format and later reformatted as jpegs with Adobe Photoshop CS3. The screenshots have been resized (maintaining aspect ratio) and/or cropped but not altered in any other way. While using jpeg does introduce a certain amount of compression artifacts to the images, we felt that they did not interfere with illustrating any of the image quality differences we are trying to point out in our direct-comparison images, where a single image displays a part of a scene in both DX9 and DX10 side-by-side in order to focus on a specific difference in image quality. After all, if the difference is so subtle that simple jpeg compression artifacts could prevent the difference from being noticed in a direct side-by-side comparison, how are you supposed to notice it in-game?
Discuss This Article