GeForce 6800 or Radeon X800? This question was on everyone’s mind this September. Drivers, applications, prices or even availability-all this has changed since the products’ release this summer and our first tests. So now is the time to take stock of the situation!
Two differing architectures We are not going go into detail again on the two different architectures. They were already explained in past articles on the X800 and 6800. However, we will remind you of the most important points.
NVIDIA introduced a new architecture with the GeForce 6800. Changes were necessary after the GeForce FX series’ limited success, mainly involving its floating calculations performance. In practice, the new architecture is a success. Compared to the GeForce FX, the GeForce 6800 is in another league thanks to pixel shaders, single texture and its 16 pipelines.
Innovations were also in terms of capabilities with a new Shader Model 3.0 generating longer shaders and adding dynamic branching. Connections are, however, costly to performance. We are waiting to see how developers are going to use this capability to know if it’s really a plus for the current graphic card generation.
The GeForce6 is also the first chip to provide full FP16 format support. It has a buffer in floating point and is also able to process blending or filtering operations to efficiently manage the HDR with flexibility. The 6800 was initially equipped with a video engine able to hardware decode WMV9 and hardware to encode MPEG1/2/4. However, this video engine as is implemented in current chips, isn’t working perfectly and fully used by the drivers. No one knows if it will be fully used one day, but for now one can consider these functions an advantage for the 6800.
There are currently three official GeForce 6800 versions: Ultra, GT and standard. Another non-official version, the LE, exists but it is exclusively dedicated to OEMs. Here are the characteristics of the three official versions:
You may notice that only the Ultra and GT have different frequencies. The official version, the 6800, has one Pixel Pipeline and one Vertex Engine less than the GT and Ultra.
Unlike the GeForce 6800, the Radeon X800 is “simply” an improved Radeon 9700/9800. This hasn’t hindered ATI because of the architecture’s excellent performance. Evolutions include a 130 nm fabrication process instead of 150 nm, a higher number of pixels pipelines, 16 instead of 8 (12 for the X800 Pro), and 6 vertex engines instead of 4.
However, two new capabilities were also introduced, that the Radeon 9700/9800 didn’t have. The first is the Shader Model 2.0b, a shader 2.0 evolution. This new shader model allows processing longer shaders (512 instructions instead of the previous 96) with ATI’s VPU. This process is used in Far Cry with the SM3.0 and SM2.0B paths. Dynamic branching isn´t compatible with the X800 because of its architecture. So, dynamic branching remains the GeForce 6800’s and Shader Model 3.0’s exclusivity.
The other innovation is the 3Dc. It is actually a compression format of “normal maps” used for bump mapping. More efficient than standard DXTC compression, this format allows it to keep a similar compression (1 for 2) and have improved quality compressed normal maps close to non-compressed normal maps. In practice, a game could benefit from the 3Dc. The compression permits a bandwidth or quality improvement. A 3Dc compressed normal map of equivalent size logically provides more detail than another non compressed normal map. Here again, only the future and developers will tell if this function is a plus for the X800.
There are three official X800 declinations with an additional X800 SE dedicated only to OEMS:
- Radeon X800 XT Platinum Edition : 16 pipelines, 520/575 MHz
- Radeon X800 XT : 16 pipelines, 500/500 MHz
- Radeon X800 Pro : 12 pipelines, 475/450 MHz
The X800 and X800 XT PE only have different frequencies. The Pro version has one Quad pixel pipeline less than the X800 and X800 XT PE. Other characteristics are identical.