IntroductionI have held off upgrading my trusty old overclocked TNT2 for quite some time, waiting till a true ‘next generation’ card came along. I have been concerned about the memory bandwidth limitations on certain of the newer cards, so the choice for me boiled down to waiting for a Kyro or 3dfx Rampage, or go for the Radeon now. Patience never was my strongpoint.
I won’t go into minute detail of the feature set of this card. I’d advise heading over to Chick’s Hardware if you want the full ins-and –outs of that. The short version follows
64 Megs DDR Ram clocked at 166mhz (only the 64 Meg ViVo version is clocked at 183 apparently):
HyperZ technology. This is a bandwidth reducing set of features aimed at improving memory bandwidth through eliminating the Z buffer bottleneck. It does this in 3 ways
1 Fast Z clear.This clears the Z buffer automatically after each frame rather than flushing it by sending a load of 0000s at it.
2 It works out which parts of the scene are going to be rendered before sending it to the Z buffer, rather than once it is in the buffer. This keeps the amount of data down
3 Lossless Z buffer compression. This squeezes the size of the data in the Z buffer without losing any of the data.
T&L engine, which also can do some neat tricks to help tidy up unsightlyjoints between.polys on models.
It has only two texture piplines, but each of these can do 3 passes. This compares with the Geforce 2s which have 4 pipelines but only do 2 at a pass. This setup probably puts the Radeon at a disadvantage at the moment as most current games don’t apply textures/lighmaps in this way, but future games may well use it to advantage.
Environmental bump mapping. Not used in a lot of games I’m afraid. Expendable and Evolva (with patch) are the only ones I can think of off hand that support it.