Not very statistically correct data but (based on only one shop ):
Komplett.se
6800: 170+ in stock
6800 LE: 100+ in stock
6800 GT 128 MB: 10-19 in stock
6800 GT 256 MB: none
6800 Ultra: none
X800: none
X800 XT: none
They have had both X800 XT's and 6800 Ultra's in stock though (and some watercooled 6800 Ultra Extremes that cost a fortune). Now, as to what conclusion you can draw from this data, well, none really
Doesn't say anything about the amount though. Anyway, i wonder if you could come to the conclusion that there's a RAM shortage that's casing the problem. That's based on the shortage of X800 XT's, X800 Pro's (available here but not at komplett.se) and Ultras combined with the availability of 128 Mb GT's and all lower end 6800's.
All based on very thorough research of course
Yeah, Bjorn, that data is more than a little ambiguous. Heck, it may be that the X800s are just selling faster than the 6800s. Or it could be that ATi has a worse distribution network in balmy Sweden, or has decided to focus more on huge OEMs (like Dell) or on other countries. Or nV may be getting better yields.
Really, it's wide open. Those 128MB GTs are weird, though. Are they a lot more than plain 6800s?
Comparing retail prices is a slightly futile task to get anything meaningful since the situations between ATI and NVIDIA are completetly different at the moment. At present, bar any remaining production of NV36, the only chip NVIDIA has on IBM's Eastfishkill fab is NV40 - off the top of my head this has one major OEM win with HP, but the rest are going to smaller OEM's, SI's and retail. While AGP X800 doesn't have many (any) Teir-1 OEM's, it is on TSMC's current most advanced (and probably most limited) process thats available for high performance graphics, and you need to look at what else ATI is producing on that process hence what wafer starts will get the priorities - everything on that process, bar X800 AGP has lots of OEM' interest so the wafer starts are going to the guaranteed OEM orders. That explains the difference in retail availability, but not potential price differences between the wafer costs.
Things to look at to understand this are whether or not there is a premium for low-k over FSG, what the die size difference are, how many layers they are using, what wafer sizes there are and whether each comapany is getting any special deals from the fabs. And I think you can go fish for any real understanding of some of those points!
Random statistics would state that there's a 50% chance that NV40 will be cheaper than R420, well within the margin of error for the Inquirer to make an (informed?) guess. In a few days they'll probably run a 'R420 Less Expensive Than NV40?' headline, just to cover all bases
No, thats written because that's what he's been told by NVIDIA. Whether or not NVIDIA has the full details of ATI's costs (beyond the default wafer costings) is another question entirely.