The last week has been very quiet in terms of updates, but behind the scenes Hermann, Brent and I have been very busy testing and reviewing. Matrox' G400 MAX, which made its way over to me much too late, will finally be reviewed by Brent Kerby. We are still collecting data, after we came across quite a few odd things. Currently, the on the first look impressive 32-bit rendering performance of G400 seems to be close to meaningless, because the actual drivers are unable to provide 32-bit deep Z-buffering. Normally, 32-bit Z-buffering is obligate for 32-bit color rendering and it is costing a significant amount of performance. Without this feature it is not too surprising that G400 is rendering extremely well under 32-bit color mode.

This was the official response from Matrox:

I am not quite sure if Matrox will be able to keep the promise that there won't be a noticeable performance difference with 32-bit Z-buffer enabled, but it's certainly true that the very fast memory interface of G400 is able to provide excellent 32-bit performance indeed. Anyway, I guess that all currently published 32-bit rendering results of G400 are pretty meaningless.

PC133

You've certainly heard about it, VIA is the first chipset maker that is able to provide a competitor to Intel's upcoming 'Camino'-chipset. Camino is meant to mainly use direct-RDRAM, which is still very expensive and seems to have questionable latency issues. The latest news say that there will be RIMM-adapters to run SDRAM on Camino-boards as well, but it doesn't seem clear what the performance hit will be. VIA's upcoming Apollo Pro chipset is using PC133 SDRAM instead and it's supposed to be competitive to Intel's Camino chipset at a much more attractive price point. We have tested boards with the VIA Apollo Pro chipset and PC133 SDRAM from Micron and Enhanced Memory Systems and the results will be made available to you this week.