Indeed this was an unprecedented 'leap' forward when 3dfx engineering pulled out all the technological stops and broke away from the rules that regulate product features and release schedules as they were not tied to Q4 financials, investors, OEM deals et al. (going public seems to have its drawbacks for the engineering annexe of any chip company). Trying to follow on from such a head start means inevitable disappointment for some of the early technology adopters, who weren't quite as excited in 1998 by 3dfx' second generation part. Voodoo2 was blazing fast (being clocked higher), packed more memory and amongst other neat stuff could handle mulitexturing in a single cycle and could be coupled with another Voodoo2 of similar spec. Still there were a 'few' skeptics that sprouted up (not many?) pointing towards the limitations of the Voodoo architecture.

Voodoo Banshee (their first single chip 2D/3D solution) was released in Q3 of 1998 and then earlier this year Voodoo3 was unleashed. By Voodoo3, however, other chip manufacturers had caught up and closed the gap on 3dfx. Although both the Voodoo Banshee and Voodoo3 were both exceptionally fast compared to their competition, nevertheless 3dfx received some criticism from some of their most ardent fans (Sharky Extreme included).

Exceptionally fast 16-bit color but with a 256x256 texture size limitation wasn't quite the leap we were hoping for. We craved 32-bit color, higher resolution textures and improved image quality features. 3dfx feels that their next product release, scheduled for Q4 of this year (you'll be able to buy it this Xmas), will curb criticism and put them back 'in front of the rest of the pack'. They're banking on their proprietary T-Buffer technology to revolutionize the look of consumer real-time 3D graphics in similar fashion to the speed splash made by Voodoo Graphics.

So what in the heck is a T-Buffer?