Yes, saying Fermi is doing well in HPC is laughable to be honest, it "launched" last October, it hasn't shipped any professional versions at all afaik has it? I've not seen a single story about the professional versions being available and thats important.
Fermi, the desktop version has its DP power stunted in drivers to around 1/4 of what the professional card can do.
It might make a large mark up on professional cards, but NOT if its not shipping any.
Likewise the R&D costs, even just the people required for support and writing the drivers for the professional versions will be higher than the profit of selling a few hundred of them.
Infact, Fermi's even at the $1500 prices, aren't making anywhere near as much profit as last gen. A professional gen card sold at $1500 last year would make $1300 profit, this time around if they were available, you'd be looking at something closer to $800 profit, they are making a loss at £450 remember.
Likewise they missed all power promises, which means a LOT of designs for big computers won't be forfilled as HPC guys care about power per watt, if they can run a 5970 which they can actually get and get more performance per watt and per $, thats what they will do. The worst part is the new TDP they've decided on isn't too unaccurate for gaming, however the same ratings put a 58xx way below the 180W rating they have aswell. 180w/5870 and 300W/480gtx are the power ratings in the likes of furmark, at 250W/480gtx and 140w/5870 TDP rating you're talking in game power usage, which just doesn't fully load the core in the same way.
However HPC work, heavy GPGPU work gets you much closer to the full "furmark" type rating, if systems are designed to use X amount of professional Fermi's at 225W a pop, people will have to completely redesign systems to use multiple 300W cards, diff psu's, cases, cooling, different power requirements for the room they were to be housed in etc, etc, its not a small job and guys that make the big setups that make Nvidia big money, take months of design and planning. Again that was if they are available, which AFAIK, they aren't at all.
The 470/480gtx prices won't settle down, its NOT IN PRODUCTION.
They do not cost £50 more than a 58xx to produce, thats utter rubbish, a wafer costs $5k, thats not under discussion anyone in the world would tell you that, if you're getting around 15 cores per wafer for both 480's and salvaged 470's, thats $5k/15= $333 just for the cores, if its less than that the price goes up.
Thats the CORE, most generations the core would cost something along the lines of $40-80 and they'd make a 100% profit and sell at $80-160.
$333 is JUST the core, its got more memory, 512mb of the fastest memory in the world costs a good $20-30 extra, probably $70-100 for the memory(prices are not good for memory of any type right now), $50 for a very complex pcb, military grade vrms/caps is anotehr $10-20, the cooler will cost $30, dvi/hdmi connectors, and putting it all together, packages with the cards, boxing and shipping, they likely cost minimum of $500, absolute bare minimum, probably closer to $600-650.
They are making a HEAVY loss on these parts, some estimates put the core count at closer to 10, if it was bang on the mark at 10 that pushes the core cost alone at $500.
Theres a reason they aren't in production they make a loss, if they were in production and were making almost the same profit, Nvidia gets a higher allocation of 40nm production than AMD do(mostly because TSMC know AMD is off VERY soon and Nvidia are going to stay long term and have to keep them happy to stop them also going to GloFo, which is a real possibility).
Remember all the most pro Nvidia sources say little to no supply now till mid/late May, and another batch MIGHT be available late in summer. If it was in production, 6 weeks after production starts you start getting DAILY wafers coming out(it takes 6 weeks start to finish to cook them at various stages). In full production means a somewhat decent quantity coming in daily to stores worldwide, thats just not happening at all.
For comparison AMD have the same wafer costs, 5k, and seem to be getting around a 50% yield on a core 60% of the size, they are getting some 75-80 cores per wafer $5k/70 = $71. The core alone is a $250+ difference in production, the pcb costs more the power circuitry costs more, the cooling costs more and the memory costs more. £50 difference, not even close. For last generation the wafers cost $3.5k and AMD were getting something like 70% yields on a smaller comparitively core, so last gen you were looking at $3.5k/110=$32.
