Eh not bad specs however in 2-3 months I'll be switching back to EVGA\GTX. I love the AMD Radeon 6870 I got now but I miss EVGA\GTX.

6.0 TFlops single precision, impressive numbers for sure. Will be an exciting Christmas no doubt.

quite possible that this is the GTX 595 killer

oh yeah the same for ATI.AMD's HD 6990 rofl
 
But as we all know a Dual GPU lacks something of the worst kind which has always been microstuttering, even that these new GPU's scale better, they will have more microstutter's than two, three or four GTX 480's or GTX 580's for example.
 
So yeah I dunno sometimes I have to admit that the GTX 480 I have now, that I am very impressed, having no more micro stuttersis surely a better thing, So I might go Dual GTX 580 or Dual HD 6970 this ride, unless the GTX 595 and HD 6990 can aprove differenty, that indeed remains to be answered, but I think it will be like any other GX2 and X2 we have seen, they all have the similar and same microstuttering issue, so likewise these new ones will suffer this as well.

Gold LeaderBut as we all know a Dual GPU lacks something of the worst kind which has always been microstuttering, even that these new GPU's scale better, they will have more microstutter's than two, three or four GTX 480's or GTX 580's for example.

 
This is incorrect. The microstuttering on a dual GPU card is exactly the same as for two single GPU cards in SLI, because they look like exactly the same thing to the driver (in fact they are literally the same hardware in a different packaging format, shared output circuitry has no impact on frame generation). Similarly the microstuttering on two dual GPU cards in SLI/Crossfire will be exactly the same as for four single GPU cards. Personally I think this is a much overblown problem anyway.

-2 Incorrect,
 
For starters the NF200 is old, outdated, obselete and should be replaced, ATi can combine a dual chip card with a single chip card, NVIDIA can't even do this yet, even that it is possible in theory, the main problem is the NF200, it's NVIDIA's Achilles Heel.
 
As for micro stuttering on dual chip cards, well to be honest I was right to a certain pioint of view, Dual chip cards have greater chances of microstutters because one PCI-Express bus has to handle two chips instead of one, so even using two Dual chip cards the chances of microstutters is greater than 4 single GPU cards.
 
Two loose cards have always shown a much healthier minimum fps than ay Dualchip card, I have a Quad SLI here and so far I find two sepperate cards better than a Dual chip sollution or two Dual chip sollutions, Quad SLI and Quad CFX  never really shown massive gains over dual chip mode as using a single GX2 or X2 based VGA.
 
The same goes with ATI setups, two singles is better than one dual of a similar class.
 
The advantage over NVIDIA dual chip cards over ATI variant , is that you can disable Dual chip mode, this feature has been dropped with ATI cards since Cat 7.10 which was last possible with R680 aka HD 3870 x2, but due to driver instability issues, this feature was left out.

when gtx280 came out it was like... OMG!!!! this is insane!!! now i wouldn't be surprised anymore lol

yellow__fever

Gold Leader

-2 Incorrect,

For starters the NF200 is old, outdated, obselete and should be replaced, ATi can combine a dual chip card with a single chip card, NVIDIA can't even do this yet, even that it is possible in theory, the main problem is the NF200, it's NVIDIA's Achilles Heel.

As for micro stuttering on dual chip cards, well to be honest I was right to a certain pioint of view, Dual chip cards have greater chances of microstutters because one PCI-Express bus has to handle two chips instead of one, so even using two Dual chip cards the chances of microstutters is greater than 4 single GPU cards.

No offense but both these statements are opinions. I have had no microstutter issues with my GTX295's that a driver update didn't fix, and although I agree that the NF200 is an annoying requirement for single GPU Quad SLI, it doesn't change the fact that it's still a requirement. Dual GPU cards in SLI with single GPU cards, although a nice feature, is (in my mind) still a configuration that isn't very popular. Since SLI is Split Frame Rendering or Alternate Frame Rendering, having a third GPU without the exact same capabilities as the first two may hinder the framerate, imagine a GTX295 SLIed with a GTX580, the 580 would be rendering frames faster than the GTX295 could which would be disasterous.

 
Well ATi stuill uses the best method which is Tiile based rendering, this delivers the best image and also makes Dualchip + Single chip possible since X1950 Dual which Sapphire made this card could be set in tri CFX with a single X1950 pro, and this is 2006 as far as I can recall, so yeah 4 years ATi has made this possible it's very popular where I come from, NVIDIA use's Split Frame Rendering and Alternate frame rendering which is something ATi used with theirrage Fury MAXX in 1999, 3dfxs way of doing is byfar still the best emthod this is Scan Line Interpleave, the only disadvantage , is that 3dfx had no advanced shaders but the performance of a quad Chip VSA-100 based VGA card such as a 3dfx Voodoo5 6000 had the same performances with FSAA x4 in games as a Dual chip card using FSAA x2 in the same titles boost from Dual chip to quad chip was always 100% and microstuttering that never took place.
 

Tile based rendering gives a better image but it does scale lesser per chip than Scalable Link interface does, which can be seen when HD5000 series  is compared to GTX 400 & 500 series
 
With ATi's method I had lesser clipping issues as renderings went when using V-Sync with CFX than with SLI, thx to Tile based Rendering, such things do not occur with CFX. CFX costs more energy to work, it works but it's less efficiant than SLI because more is ging on with the cards, workload of SLI is lesser than CFX m thus the workload is less , so this is why SLI also scales better, thx to my research over the years I did find this out quite soon, which was a few years back, 2006 it was somewhere.
 
It's is very interesting to see that only new-gen Dual chip cards lack from this micro stuttering problem, what trigger's it, is still unknown, no one could really prove this as my VGA card collecting and research days go.
I have had alot of VGA's you can see them here:

Click to Enter
 
The ones from the previous collections are card I don't have anymore but I rememebr working with them nione the less, the Previous 3dfx collection map holds a lot of 3dfx Dualchip and quad chip card photo's I made when I had them and tested them , feel free to have a good look around, i ahd to sell them because of health issues and some depressive syndrome that took place, such things always happen at the wrong time.
 
To combine a GTX 595 with a GTX 580 is what I'd wnat, this would be more interesting and cost effective than two of them cards and it would scale better as well, ATi can do this a HD 5970 + HD 5870 same idea or HD 6990 + HD 6970 :)
 
You can also CFX a HD 4870 X2 with a HD 4890 or a HD 5830 + HD 5870 + HD 5970, with ATI you can go more ways with NVIDIA the host card has to be exactly identical as with with apired one the slave card that is, 3dfx also had this with Voodoo2's if you had a 12MB card your other V2 had to have 12MB also, however there have been experiements that combining am 8MB card with a 12 MB card was possible due to some driver hacks but this always lead to unstable performance., with ATI ram size has no issue, all what will happen is that the card with the most less amount of ram will take the upperhand over the one with the greatest amount.
 
Example have 1x HD 4870 512MB with 1x HD 4890 1024 MB + 1x HD 4850 1024MB all cards will have a ram buffer of onlu 512MB leaving the other ram half unused, even so all 3 VPU's will be uused in tri CFX, it's always better to have more VPU's for some appsthan more ram, ATi did a good fine job on this, somethingwe won't see NVIDIA doing, they don't have enough experience, ATI has been doing multi chip since 1999, NVIDA has been doing this since 2004, as for 3dfx they have been doing this since 1995.

As for the option to go single chip,well simple even today there are plenty apps that don't support dualchip, so by disabling one, you can save performance loss as well, on the otherhand you can also use a dual chip card by disabling 1 GPU and use it as a detiocated PhysX PPU instead, this is a handy setting for the ones with only one PCI-E slot, this is why i find it important to have that option always available, since it has multiple reasons to be there in the first place
 
And  don't worry I am not offended that easilly

it all depends ATi's scaling has improved again in their 6000 series which is only a refresh, their HD 7000 series orsay their Southern Ilands series will be a good compettition for NVIDIA's Kepler, and somehow ATi might even beat them as well, HD 6000 series proved much better scaling than HD 5000 series and then that two mid end cards like the HD 6870 can keep up with 2-way SLIéd GTX 480 and sometiems 580 in a few games was very surprising.
 
HD 6970 will launch soon, so yeah some cases NVIDIA will have the upperhand and with others ATi will, I think this will be like the 6800 Ultra and X800 XT PE, two very similar performing VGA cards.
 
ATi uses Tile based rendering which results better image quality NVIDIA uses Alternate Frame rendering which results better GPU to GPU scaling, but overall ATi has improved scaling wise again.s o they might even beat NV fromt he throne here, I have a feeling that something biig will happen when the HD 7000 series will have to battle against NV's GTX 600 series, this is going to be interesting none the less.
 
All we can do is observe, afterall they are just VGA cards to me and are just fun to collect, compare and to explore.