I am now confident I'll wait for the nvidia GTX, if 2 of their mainstream gpu's can be cheaper and beat the best Ati card (2 gpu) by 8.15 % in real games.

http://en.expreview.com/?p=273&page=16

" We’ve only run two games and one 3Dmark06, but the score is very weird. While Hd 3870 X2 leads in 3Dmark06, it lose to 9600GT SLI in both call of duty 4 and Bioshock."

It could be a driver problem but anyway the Ati X2 should have the advantage since it's been around longer.

Anyway this seems very promising for the new nvidia high-end cards and SLI scaling, I hope nvidia will suprise us and show that all the specs of the GTX we got till now are fake.

Looks promasing for Nvidiabut i will wait for a review from someone i have heard of and one that is conducted on a full set of benchmarks before i take it as gospel/
Mactronix

not so great, tbh... a dual 9600gt (w/e it's called) is the same solutiion as a 3870X2...

and 8.16 percent is not a lot... if it's cheaper then fine, but if not, then forget it, i'd be happier with an X2 solution, which can take advantage of Xfire's better GPU scaling....

Why is everybody always saying that crossfire scales better?
In most crossfire vs sli reviews that I read SLI scaled just as good...
It seems to depend on the game and res and frame quality.
The thing that always comes back is that scaling of the second gpu is good.
But the 3th and cerntainly the 4th did not seem worth it.
And this should become even worse with DX10 games due to the way it works.
So imo better have 2 high-end single card GPU's than 3 or 4 others.

Anyone know how crossfire works compared to SLI?

gpu - northbridge - CPU - northbrigde - GPU ?

Or does it work like the new enhanced SLI on the new 790i :

gpu - northbridge - GPU

The last time I checked, the 8800GT SLI outperforms the 3870x2 as well. What is the big scurry about?

http://www.anandtech.com/video/sho [...] i=3209&p=4

http://www.firingsquad.com/hardwar [...] /page3.asp

That comparison is lame, because the 3870x2 has not been out that long, and Nvidia's been known to fudge quality on pre release drivers for extra fps, but to correct image quality issues after the first benchies. Therefore, drivers for either card are not mature.

I'm not defending the 3870x2 against the upcoming 9800gx2, because I believe the Nvidia dual GPU card will win until the arrival of the 4870x2 ups the ante, but it's silly to compare a low end SLI that no one would buy to a viable dual GPU because they're aimed at different markets. I'd say the same thing if the upcoming 4850 in Crossfire ends up beating the 9800gx2.

IMHO, anyone building a new system, trying to decide between Crossfire and SLI would want to see how high end cards, especially dual GPU, compare in CrossfireX and Triple SLI. On the other hand, anyone with an existing SLI board an an older 8 series card like the 8800gts 320 might want to know if two 9600gt's are a deal.

No one building a new system would go 9600gt in SLI. They'd be better off spending that additional $50 for a 9800gx2 or a single 9800gtx for reasonable SLI performance down the line. An apples to apples comparison between ATI and Nvidia involving the 9600gt would be a single card test including the 3850, the 3870 and the 8800gt (both models).

In those tests, it's been a mixed bag with some game benchmark wins for Nvidia and others for ATI. In those tests, it wasn't just between a couple FPS, but involved RPGs and RTS titles. Those are the ones that I trust regarding 9600gt performance.

You say no one would go 9600GT SLI, however it is a good option for those who want a good price/performance system at this time. The pricing of the 9800 series is unknown at the moment and could drastically change from estimates in relation to how the supplies pan out. Those working with a tight budget may not want to dump that much $$ into a SLI GTX system or wait around for pricing to level out.

IMO the GX2 has lost it's mojo. No one quite knows why SLI has improved with the 9600GT at the moment but if it translates well to the GTX, the GX2 won't really have a place to stand out. I think they should just toss it in the same waste bin with the 7 series x2 cards.

1. Because it does prior to the 9600GT

2. What reviews? Post results to support your statement.

Well yes, but the 9600GT is a midrange card.

Funnier still the HD3850 destroys the GF9600GT;

http://en.expreview.com/2008/02/23 [...] w/?page=10




OMG, the HD3850 roxors and the GF9600GT & HD3870X2 SuXorZ!

Now you might say that's selective testing, but hey, it's their tests and their selection of tests, I'm just giving value to some tests over others, kinda similar to the small glimpse their review shows, the InQ picked up on and then you repeated here.

Personally that page you link to (and that the InQ picked up on) with it's cherry-picked tests shows me more about expreview than either of the cards/solutions.

As has been shown in pretty much every review sofar, the GF9600Gt is a great card, but like all mid-range cards, their performance looks very impressive in weak titles, but increase the shader workload and they fall away. Why would you test the less graphically intense games for an Xfire/SLi test? Hmmm....

Ape has the right idea here!
You can prove anything you like with numbers...
Even if it does perform better than a 3870X2 I'm hardly going to chuck in my GTX for two inferior cards am I?
That and as yipsl says, nobody is going to build a new PC with 2 9600's...
A good midrange card, but I can't help but feel it doesn't have space in the marketplace, between the 3850/3870/8800GS/8000GT....

Well that's the FIRST TIME ever that I've seen somebody try to use a HIGH END card to knock down a MID-END card!

Seriously, that's just like saying the 8800GT demolishes the HD2600XT. Except .... The 9600GT isnt being demolished AT all...

ATi's HD3800 series are weak cards, they shouldve branded them HD3600... and branded the HD3600's as HD3470s...

The HD3800 name shouldve been reserved for something better performing & higher priced. It just makes the 3 series cards look poor compared to the 9 series ones and average compared to the 8 series cards.

All I wanted to say is that the current reviews of the 9600GT make the new GTX (in SLI) look very promissing.
I was not trying to say we should buy 9600GT SLI, I'm hoping to get the new GTX SLI if it lives up to my expectations.
I'm fully aware that most reviews should be read with a pinch of salt.

I'm not going to look up all the reviews I read in the past about SLI but here is one of the latest. At the end of the page :
"Single to dual GPU scaling is similar with SLI as CrossFire, but the 2-way to 3-way gain is better on CrossFireX."

http://www.anandtech.com/video/sho [...] i=3232&p=3

I'm not trying to prove anything of convince anyone, I'm just trying to share my toughts. I love to read different opinions and learn new things. And I admit I sometimes have to change my mind after reading them. Anyway I'm happy to read all the reply's :-)

No one notices how poorly crossfirex scales with those two games? (CoD4 and Bioshock) take a look at the Page 11: 9600 GT vs Radeon HD 3870 and compare those to the 3870x2 ones. I think thats just bad drivers

Time out. They never tested xfire with the 3870. You can't judge xfire scaling with the comparison of a single 3870 to the 3870x2. Is that what you are saying or am I reading your post wrong? Plus they aren't using xfirex either.

Take this cherry picked benchy for example:




Now compare to this one and you have nothing to go off.



Haha, actually this one is better: