I don't blame ATI simply because they have no reason to drop prices until there are competitive single GPU alternatives to their high end cards with comparable price/performance. And that's even if yield ratios weren't lower as they continue to improve the 40nm process production.

When you add to that the fact that many vendors can't keep those high end cards in stock, expecting a price drop would have been like expecting a price drop on a Nintendo Wii the first couple years of production when they were rarely found on shelves.

ATI simply has no reason to drop price until output meets demand. It usually happens faster, but with Nvidia's delays on their current gen GPUs (and now potentially with a product that does not fully compete with ATI's best), ATI can and frankly should take advantage of the situation to help AMD's bottom line.

Actually I've just noticed on Newegg a lot of the cards have had a slight reduction in price. The HD 57xx series are starting to get incentives like mail in rebates applied and the whole of the range is edging lower in price. The lowest 5850 is now $290 and the lowest 5870 is now $395 which is about a $10-15 cut in price from the highest point. I suspect this is a function of supply bringing the prices down to match the launch prices rather than any concious effort to compete with a non-existant card.

The MSRP, if something even exists for such price elastic items as VGA cards, is probably still about $299 and $399 for the 5850 and 5870. Initial reviews for those cards had street prices around $20-30 below those prices (Anandtech listed as $379 and $259), probably accounting for the typical discounting expected. It wasn't until output and supplies undercut demand that we actually saw prices spiking above the base line $299/$399 prices.

Naturally, they can't stay static for too long (they are VGA cards), but it just might take longer than usual before we start to see the normal heavy discounts and deals, or we see ATI drop prices on those cards to vendors.

But it's already been about 5 months since they debuted which is an almost unprecedented length of time before seeing such discounts.

   This "article" is published by Charlie the biggest anti-nvidia troll in the whole internet.

It's like trying to find resonable x360 game review on SDF :D

Does that have any bearing on whether he is right? He's made predictions, they're not ambiguous and they can be proven or disproven when Fermi launches.

Unless you can point to specific claims on Fermi where he's wrong? Because he's been saying the same stuff since last May, when everyone else thought this was a crazy position to take.

Do I need more proof than this article title to prove he is biased?

http://www.semiaccurate.com/2009/07/29/miracles-happen-gt300-tapes-out/

Seriously he puts common internet knowledge puts some secret "sources" into mix and try to pretend he has any information when at most he is making worst case scenario educated guess.

Yeah, ignore the unfounded sensationalism. If you read his forum comments you'll see none of that (he's actually a nice guy), it's just to get readers. Just work with the facts and details. But what specific facts are you disputing? As in, precise claims about clocks or power or release date? I can't see where he's wrong yet.

Really, say what specifics you think are wrong. I can't argue against a strawman.

His GT300 tape out date was actually, at the time, very different from 'common internet knowledge'. The consensus was around BSN's date of January to March tapeout. Who was right? Charlie. If they'd actually had silicon in March we would have seen chips by now. Whereas Charlie has given precise dates and rationales for the A2 and A3 tapeouts too which are all consistent with the lack of chips and the release date of this March.

Here is a good article about the situation as it is now. It is about ATI's RV870 chip but reading through the article (and occasionally between the lines) one can figure out the (serious) problems NVidia is currently facing: 

http://www.anandtech.com/video/showdoc.aspx?i=3740

On the other hand, I find it rather strange that people think an entire fermi/RV870 based graphics card (chips at least as complex as the best processors available) should be available for a measly $199-ä299.

That article is really good, I'd recommend it to anyone. It's the story of RV870's creation straight from the engineers who designed it. I'd also recommend the article written for RV770.

The 5850 and 5870 are priced a little high compared with the previous generation. But that is to be expected given the lack of competition and the fact demand still exceeds supply (Proof: look at the 5970's stock situation). They're still rationing Cypress chips, so they can charge what the market will bear.

Compared to 2007 and earlier though, these prices are fantastic. Remember when top-of-the-line was $800 with the 8800 Ultra? Or how nothing under $250 was a decent gaming card with X1x00 and 7x00 cards?

Charlie may be a troll sometimes, but he's been right more often than not with regards to Nvidia and Fermi.