Synthetic Direct3D 10 Tests of
AMD and NVIDIA Graphics Cards
in RightMark3D 2.0
Quite some time has passed since the launch of Microsoft Windows Vista and the updated DirectX 10 API. There gradually appear games supporting Direct3D 10. They are just modified D3D 9 applications, and don't actively use new features of D3D 10. And the existing games surprise us with opposite results of graphics cards from the two main chipmakers. Of course, we can use them for comparisons, but it should be done with care...
We don't see many synthetic and game tests supporting Direct3D 10 either. For example, Futuremark hasn't released the next version of 3DMark. But iXBT.com has its own synthetic benchmark, and we've been planning on updating RightMark3D for a long time already, in order to evaluate peak performance of D3D10 graphics cards in various tasks. The final version of RightMark3D 2.0 for benchmarking Direct3D 10 compatible graphics cards in MS Windows Vista has been released only recently. And we've immediately started using it in our articles.
Some previously known tests were rewritten for DX10, new types of synthetic tests are added: modified tests of pixel shaders rewritten for SM 4.0, tests of geometry shaders, vertex texture fetch tests. It will be the first article about RightMark3D 2.0. We've tested a lot of graphics cards here. From now on we'll use the new benchmark in our articles.
It should be noted that all these tests are purely synthetic. They usually use only certain GPU units, while influence of other units is intentionally reduced to minimum. All these tests load GPU only, CPU is not used. Now let's proceed to the long-anticipated DirectX 10 tests!
Test conditions
Our version of RightMark3D 2.0 and its brief description can be downloaded at the following link: http://www.ixbt.com/video/itogi-video/ini/rmdx10.rar (4.5 MB)
RightMark3D 2.0 requires MS Visual Studio 2005 runtime as well as the latest update of DirectX runtime.
Note: the screenshot shows a special version of RightMark3D 2.0 that can test graphics cards in the batch mode. It's developed for our in-company tests, and we'll release it a bit later.
Testbed configuration:
- Intel Core 2 Duo (Socket 775) based computer
- CPU: Intel Core 2 Duo Extreme X6800 (2930 MHz) (L2=4096K)
- Motherboard: EVGA nForce 680i SLI on NVIDIA nForce 680i
- RAM: 2 GB DDR2 SDRAM Corsair 1142MHz (CAS (tCL)=5; RAS to CAS delay (tRCD)=5; Row Precharge (tRP)=5; tRAS=15)
- HDD: WD Caviar SE WD1600JD 160GB SATA
- PSU: Tagan 1100-U95 (1100W).
- OS: Windows Vista Ultimate 32-bit; DirectX 10
- Monitor: Dell 3007WFP (30").
- Drivers: ATI CATALYST 8.3891; NVIDIA ForceWare 158.45.
Synthetic tests were run with the following graphics cards:
- RADEON HD 2900 XT with standard parameters
- RADEON HD 2600 XT with standard parameters
- RADEON HD 2600 PRO with standard parameters (with GDDR3 memory)
- RADEON HD 2400 XT with standard parameters
- GeForce 8800 Ultra with standard parameters
- GeForce 8800 GTX with standard parameters
- GeForce 8800 GTS with standard parameters (modifications with 320 MB and 640 MB of video memory demonstrate similar performance)
- GeForce 8600 GTS with standard parameters
- GeForce 8600 GT with standard parameters
- GeForce 8500 GT with standard parameters
In our comparison we'll use AMD-NVIDIA pairs of graphics cards with similar positioning: HD2900XT — GF8800GTS, HD2600XT — GF8600GT, HD2600PRO — GF8500GT. Some of the graphics cards are not yet available on the market, and if their real prices are different, you should adjust our conclusions. Prices are changing constantly, many of our conclusions are true only for the time this article is published. It has nothing to do with the theoretical comparison of Low-End and top GPUs from the same company, evaluation of their relative performance does not depend on pricing.
