Are 2x 8800 ultras > GTX 295?

I want to upgrade my graphics for gaming but I don't know what the best route is. I'm currently running SLi with evga 8800 ultras on my rig.

Will upgrading to one GTX 295 give me better performance? Or should I wait until a newer nvidia series? Which series should I look out for better performance?

yess a GTX295 would give you better performance.

The new nvidia cards are soon to be released, so you might want to wait until then.

You could buy a HD5870 with roughly the same performance as a GTX295 and costs less, has DX11 support, uses less power and produces less heat vs. the GTX295.

OR if you have deep pockets you could opt. for the 5970.

With the specs that we already have ATM, it will be better performance-wise, price-wise not so much, the 470 will be priced at around 499.99$ and the 480 at around the same pricepoint as the 5970. The problem that AMD has is that the 5830 and 5970 are having issues and if they are not fixed before Fermi's launch it will hurt AMD when it comes to sales. The issue that Nvidia has is that Fermi will be sold in limited quantities, meaning if you are not able to get your hands on one once released, you will eventually see them overpriced on e-bay

What was mentioned originally was EVERYONE would have 'better' OPTIONS when Fermi comes out, its you thats getting hostile at seeing the word better and Fermi in the same sentence.

Yeah, that what was being talked about until OvrClkr started talking crap about something that is unknown. It might well be better, but then again.... it might not!!! Unless OvrClkr has some new facts (in which case I appologise!!) then he does not know what he is talking about.
I wouldn't be stupid enough to make such a statment unless I knew that was the truth and could back it up with evidence (from an independant and reliable source).

Nobody knows how Fermi will perform, I believe the Fermi (I presume you mean the 480) will be quicker than a 5870 but I wouldn't want to start talking about it as if it were fact... UNTIL IT IS FACT!

You have no clue as to what we already know, stop being a troll and use some common sense.

GT400 "Fermi" Architecture GPU Specifications :

3.0+ billion transistors
40nm GPU by TSMC
384-bit memory interface (6x64-bit memory controllers)
512 shader cores (renamed to CUDA Cores)
32 CUDA cores per shader cluster
1MB L1 cache memory [divided into 16KB Cache - Shared Memory]
768KB L2 unified cache memory
Up to 6GB GDDR5 memory (1.5GB for GeForce and up to 6GB for Quadro/Tesla)
Half Speed IEEE 754 Double Precision
16 Streaming Multiprocessors (new name for the former Shader Cluster) containing 32 cores each

That right there^^ tells you that it will be superior to the 5870 , and I am not "talking crap" about anything.. I am just posting what we already know..

Nvidia is not going to sell a GPU slower than the 5870 at a much higher price, it aint happening

The specs for the 470 are mentioned above, that tells you that it will be faster than the 5870 at around 500.00$, the 480 specs should be a bit higher since it will be priced at over 650.00$ . Of course it is virtually impossible for a single chip GPU (480) to beat a dual chip GPU (5970). I am one that is waiting on the 470, once I get my hands on one I will test it out and post some numbers

Again use a bit of comon sense....

The 5870 and GTX 295 are on par performance-wise (give or take) correct? Well then that means that the 470 and 480 HAVE to be faster than the 5870 because Nvidia is not going to produce a GPU slower than the 295 at a much higher pricepoint

When figuring out what Nvidia would "dare" to do have you taken into account that the GTX 285 is about $100 more expensive than the HD5850?

You are correct , now lets see how many 5850's have been sold vs. the 285's since the launch of the 5 series . I don't think Nvidia is looking to overprice and underperform once Fermi is released, on the contrary we should see very good performance at a high cost..

All that is fine and good, but if Nvidia can't hit their clockspeed targets, then they're stuck with whatever they can get, whether or not it is faster than the 5870. My guess is that it will be faster than the 5870, but probably not by very much. Nobody will know anything for sure until they release though.

All cards don't scale the same amount in crossfire/sli. Also a bit more powerful than the GTX 280 is more like an HD5850 than an HD5870.

Yeah your not gonna want to upgraid to gtx 295 I say wait

Did you REALLY use the hierarchy chart to try to prove a point?

Again the hierarchy charts are not very helpful in these circumstances. I have fixed the typo, 5850 performance was my intent.

As you can see the 8800 Ultra is a little below a 3870 X2 (the 8800 GTX used is likely at similar clockspeeds), which is close to the 4870 512mb. 2 4870s 512mb where about 10% slower than the 1GB versions used in the 4870 X2. This puts dual 8800 Ultras at about in between the 5850 and the 5870, which is too close to the GTX 295 to be worth the upgrade.

Going by this then the 5970 is about 40% faster than your current setup which is a pretty decent upgrade:

STFU, specifications mean nothing. How will it be superior to the 5870?
Since you didn't specify I'll use an off-hand one.
2 words why 5870's architecture is superior: clean code.
Clean code aims to make code as clear & readable as possible by:
As few dependencies as possible
All dependencies are grouped together
Code is written according to the language's published standards, strictly conforming
The style of the code if uniform throughout the application

By having few & close dependencies that means the 5-SP blocks can process at peak efficiency (5 or more parallel instructions per clock cycle per thread).
This means, by making good code, you are inherently optimizing for the 5-SP block arch (R600, RV670, RV770, RV870).
So if a very parallel app (F@H, encryption tools, transcoding) is used on the GPU, the 5870 will be 1600 SP's vs Fermi's 512.
That's 3.125 times more performance if there's no clock differences, memory differences, and everything is simple instructions.

Everything comes down to context & code.

nVidia's SPs do 4 times the work of a single ATI SP meaning Fermi will have an equivalent value of 2048 of ATI's SPs.

In OvrClkr's defense, the specs that we are almost sure about tell us that Fermi should be in between the 5870 and the 5970. That said we don't know something worrying and important, clock speeds.

Well just FYI when I typed that, it was in a very low tone of voice , i really meant stop being "ignorant"

anyways I am 99% convinced that it will be more powerfull, of course drivers will be a PITA and we wont really see its full potential till they mature..

Hold your horses son, Im just giving my opinion on what "I" am expecting. That should not bother you at all since you are 100% convinced that the 5870 will trounce Fermi . One month from now we will see

Yea I agree completely, it will be a major fail for NV if the 470 comes out to be slower than the 5870, hopefully the 470 can equal or better the 5870 and the 480 should be around 10/15% slower than the 5970. Of course we all know that there is no way that the 480 will be able to give the 5970 a run for its money. NV's next dual GPU should be in the 800.00/900.00$ arena and by then the 6 series (28nm) will be mainstream....