Though the RADEON VE is clocked at 183MHz (similar to ATI's flagship RADEON product), several features had to be cut out of the 3D pipeline to facilitate the addition of the VE's other unique capabilities. Most significantly, the second pipeline that comprised the Charisma Engine was eliminated, taking three texture units along with it. There are both positive and negative ramifications of this decision. First, with only one pipeline, the VE has a theoretical fillrate of 550Mtexels/s. However, since today's applications generally offer support for two textures to be processed simultaneously, the third will often sit idle, creating a realistic fillrate of around 366Mtexels/s. On the other hand, the more-simplified 3D pipeline has allowed ATI to utilize a passive heatsink, which is valuable in the never-ending quest to eradicate fan-induced noise in the corporate environment.

Another omission from the VE is hardware accelerated transform and lighting. Unfortunately, we have found T&L to be a useful feature in low-resolution game play, which is where we expect RADEON VE owners to play. It should conversely be noted, though, that as processors become faster and drop in price, T&L might never even be missed. Four-matrix skinning and keyframe interpolation also missed the roster, making the VE less than ideal for upcoming DirectX 8 applications with support for these new features.

The good news is that the quality-enhancing features of the RADEON made it into the silicon. For example, all three types of bump mapping are supported in addition to environmental mapping. Both DirectX and S3 texture compression are support alongside shadow mapping and range-based fog.

In a measure to keep board costs down, ATI is using 32MB of 5.5ns DDR SDRAM clocked at 183MHz. The difference between it and the other members of the RADEON family is a 64-bit memory bus. Moving away from the 128-bit bus allows ATI to design a board with fewer traces and still use DDR memory to obtain acceptable bandwidth. Most cards based around NVIDIA's GeForce2 MX utilize SDRAM on a 128-bit bus, which generally benefits from more efficient data transfer due to some inherent latencies with DDR memory.

In order to combat some of these losses, ATI has also equipped the RADEON VE with the Fast Z Clear portion of the HyperZ technology. According to ATI, the Z clear feature is the most effective bandwidth-saving portion of HyperZ, and helps the RADEON core overcome the limited memory pipe associated with a 64-bit data path.

Probably the most recognized proficiency of ATI's product line has been exceptional video support. Thankfully, all of the video features from the RADEON family have carried over to the VE, making it the top-performer among the competition.

Motion compensation, iDCT, adaptive de-interlacing and alpha subpicture blending are all included and supported by the Multimedia Center 7 software. For more information on ATI's video features, refer to our RADEON Guide.