it would be an amazing piece of engineering dont get me wrong, and if someone pulled it off in a decent way it would be awesome if that same cooling tech was carried over to everything else, but its gonna have to be something fairly extreme. Ace: the GTX295 was 2 GTX275's, and those chips didnt pull 280w each - just remember that the card as a whole has to pull less than 300w to pass pci-e certification, and im not sure that any dual-fermi card would ever do that.
well exactly - but then you get into the law of diminishing returns. does the HD5970 represent a good performance/cost? did the ASUS Mars? i just re-read the CPC review of the Mars. An excellent feat of engineering, but the fact it ran much much slower than it should have tells the story.
I've said it once and I'll say it again - until we see a real-world benchie of these cards we don't really know. So far it's pretty obvious that a) it's a big core b) it might be pretty toasty given the cooling holes for a big fan at that end.
So it might double up as a stove? I've always wanted to fry some eggs and bacon while I'm gaming. I can get peckish at times.
amd is prepping 28nm gpu's for 2011.. so nvidia is not far behind once they get these out.. fermi will probably be pretty good on a 28nm die given what we know about the 8+6 power connection so far you know this thing is going to run hot as hell.. I really want to see the numbers (from a site like this one preferably)
nvidia is already a generation behind when it comes to process tech - AMD used the HD4770 as the test which is why they had relatively few problems with the 5000 series. nvidia didnt make a full-size chip on 40nm before fermi which is partly why its so late.
I can confirm these pics are legit (I was desperate to source them myself), but clock speeds, throughput etc is still a mystery. We should be getting our review samples soon though, so much excitement ahoy.
lol at least nvidia made a good vid about it was it really that bad? and why did thery make it so bad?