At this point you may be wondering about this HyperZ technology that increases both fillrate and bandwidth. We have been told that ATI's patented technology utilizes a number of techniques to eliminate bandwidth usage of the card's Z-buffer. What are the results? In a best-case scenario, the fillrate jumps from 1.2 to 1.5 Gtexels/s and the bandwidth is increased from 6.4GB/s to 8GB/s. We're not sure if this technique of obtaining "free" performance will have any negative implications, so judgment will have to be reserved until we receive a card to test.

As games become more complex and textures increase in detail, larger frame buffers become necessary to facilitate the space required. However, contrary to popular belief, size isn't always the most important factor. In order to move the information around as quickly as possible, greater memory bandwidth is needed (especially at high resolutions and 32-bit color). This bandwidth is achieved though wider memory busses, RAM that runs at higher frequencies and lower chip latencies.

With the MAXX technology, ATI has already established themselves as the industry leader in providing memory bandwidth, allowing respectable performance in high resolution, 32-bit color. The Radeon 256 will not only surpass this feat, but will initially do it with a single chip solution.

ATI has wisely chosen to support both SDR and DDR memory in configurations of up to 128MB. At up to 200MHz, the card's theoretical bandwidth will peak at around 8GB/s. Will all of this bandwidth be used immediately? Probably not, as the bandwidth of today's DDR-equipped cards aren't being pushed until 1280x1024, but as the features of the Charisma engine are exploited in the future, super-high bandwidth will be crucial.