Tom heeft een artikeltje in elkaar gedraaid over het Intel Developers Forum dat afgelopen week plaats vond. Naast info over Coppermine, Merced, USB 2.0, Geyserville mobile CPU's en Bluetooth heeft-ie ook een stukje over de GeForce 256 (die vorige week door nVidia op het IDF werd ge´ntroduceerd):
Anyhow, the demonstration went on to give several graphical examples of what the GeForce could do. We were shown models of a tree, fire truck, sports car, bubble of water, boat on a lake and a cyber woman. Most of the models had a with high poly count sample and toggled into an insane poly count version. Each time the changes were so dramatic that it blew my mind. The detail that high poly counts offer is beautiful. The scenes were all gorgeous with fine details on just about everything. However, they claim to be able to sustain a 60 FPS average while using these high poly count scenes and I did NOT see that. I assume that this might be due to several possible reasons (it was being projected, early drivers, overly poly intensive demos) but you can be sure that Tom's Hardware will have the scoop very soon. The reason NVIDIA is able to bring us this insane performance is due to the integrated T+L engine. This makes lighting and transformation VERY fast and allows the graphics card to put mind-boggling amounts of polygons on our screen. The power of a SIMD processor running at 500 MHz might do something like 2 GFLOPs for example while NVIDIA claims that a Cray super computer would do around 38 GFLOPs. The GEFORCE 256 does a whopping 50 GFLOPs.
Mooi, dan hebben wij 50 super computers .