• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

2 4870x2 crossfire for big performance boost?

I think tri fire with another 4870 may be a better choice, I've read that scaling isn't too great when going to 4 cards.
 
no. Its just not worth the extra heat and noise when most games wount even take advantage of it.
 
Its not really worth it. I mean technically for $100 a 4870x2 is a pretty sweet deal. But I remember seeing that scaling with 4 cards was not as consistent as with 3 cards.
 
I'd say go for it, the scaling may not be ideal but that much GPU performance for $100 can't be beat, and then you can always use it in another system if needed.
 
$100!!:eek: Where? Cause if you don't want it, I sure as hell would buy it for a 100 bucks.
 
Nah, you'll need a new PSU to run two 4870X2s with an i7, because that 750W isn't enough. $100 for a 4870X2 is good (and very rare) but $100 + the PSU doesn't make it worth it. May as well sell the first 4870X2 and get something more efficient. Your net cost for a 900W PSU and that $100 card will be nearly $300. Your net cost if you sell your existing X2 and buy two 6950s will only be about $400, and will be quite considerably faster, especially on a 24", where quadCF doesn't work that well.
 
There is simply no reason to do that.

4870x2 is a Dx9 card, capable of insane levels of AA at 2560x1600 in most Dx9 games not named Crysis.

Assuming optimal scaling, you would get double the frames, at double the heat.

But with a 4870x2 you are already having excellent frames in most games anyway, especially if you are on 19x12 resolutions.

And if you are at 25x16 and want better AA, the 1Gb per GPU simply wont cut.

So if you have the money, sell the 4870X2 ASAP and grab a 2Gb 6950, before AMD make only unlockable 6950.

The 2GB 6950 is a great card , more than capable of taking over the 4870X2, the unlock feature notwithstading.

Now, if you are a lucky guy with a 4x D-DVI 4870x2, like the ASUS one, well, keep it for business projects. it is a stock market player dream come true, capable of driving 4 30 inches monitors alone.


Take heed of the heat and power problem. Much was said about the heating effect on the ambient with fermi cards,

But SLI Fermis is 50 w less than Quad Fired 4870x2s. You will have to increase air conditioning in the PC room, unless you live in North Canada.:p
 
There is simply no reason to do that.

4870x2 is a Dx9 card, capable of insane levels of AA at 2560x1600 in most Dx9 games not named Crysis.

Assuming optimal scaling, you would get double the frames, at double the heat.

But with a 4870x2 you are already having excellent frames in most games anyway, especially if you are on 19x12 resolutions.

And if you are at 25x16 and want better AA, the 1Gb per GPU simply wont cut.

So if you have the money, sell the 4870X2 ASAP and grab a 2Gb 6950, before AMD make only unlockable 6950.

The 2GB 6950 is a great card , more than capable of taking over the 4870X2, the unlock feature notwithstading.

Now, if you are a lucky guy with a 4x D-DVI 4870x2, like the ASUS one, well, keep it for business projects. it is a stock market player dream come true, capable of driving 4 30 inches monitors alone.


Take heed of the heat and power problem. Much was said about the heating effect on the ambient with fermi cards,

But SLI Fermis is 50 w less than Quad Fired 4870x2s. You will have to increase air conditioning in the PC room, unless you live in North Canada.:p

The 4870x2 (and all other 3xxx+4xxx cards) are DX10. Unless DX11 is a requirement, that shouldn't be holding him back.
 
This isn't meant as a knock against any particular individual on this forum, but most of the people who talk about bad scaling with 3 or 4 GPUs have never used one. At best, they might have "read something" a few years ago that said scaling was bad. Of course, you can't expect 100% scaling.

I use this setup right now and scaling is actually very good, and so is overall compatibility. With crossfire application profiles now separate from the drivers themselves, things have never been better.

I'm not trying to make it seem better than it is, or trying to say it would be faster than a newer setup, but I don't think you would do better with any other card for ~$100. I was originally going to go with a 5870 when I got mine, and ended up getting both cards for less than a 5870 was going for at the time. Considering their performance, these cards really do offer amazing performance for the price on the used market.

Others are correct however that your power-supply will very likely need to be upgraded. Here is what my power consumption is like with my 4.4Ghz Q9650:

powerkaw.jpg
 
DX10.1 actually. Even the HD2 and HD3 series were DX10!

GotNoRice: I agree. Scaling was actually quite good on my pair of 4870X2s (Not as good as with my pair of 6970s though, I'll admit!).
Holy crap at the power usage though, is that furmark? This is what I scored, with a 3.7Ghz Q9550:

Idle (desktop): 305W
3D clocks (desktop, alt-tab from game): 450W
Light load (Heroes of Newerth): 570W
Heavy load (Left 4 Dead 2, 3dmark06 Canyon Flight): 830W
OCCT GPU stress test: 1040W
OCCT PSU stress test: 1120W

and here's my 6970s:
Idle (desktop): 160W
500mhz clocks (desktop, youtube): 195W
3D clocks (desktop, recent alt-tab): 220W
Light load (GZDoom): 290W
Medium load (Heroes of Newerth): 430W
Heavy load (Left 4 Dead 2): 530W
OCCT GPU stress test: 610W
OCCT PSU stress test: 720W
 
Holy crap at the power usage though, is that furmark? This is what I scored, with a 3.7Ghz Q9550:

That was just while playing Bad Company 2. I ran furmark (actually MSI kombustor) once but power consumption almost hit 1200 watts and I chickened out :p
 
That CPU must be volted up pretty high then, because even at 110V your PSU isn't much less efficient than mine.
 
sorry, my mistake. Its a Dx 10.1 card.

But there was only one title for Dx 10.1 Assasins Creed , and that was for a short time- LOL

Anyway my point stands true: There are not a ton of Dx 10.1 games out there that are unplayable with a single 4870x2 and all of a sudden achieve playable frame rates going quad fire
 
HAWX was DX10.1 too. DX10.1 was a farse, but its inclusion in DX11 is a lot of the reason why DX11 is less demanding than DX10 in simpler titles, it's much more optimised.
DX10 and DX9 is a fairly substantial difference. DX11 didn't bring that much new to the table.
All I can say is at max detail 2560x1600 in Left 4 Dead 2, the game was noticeably non-fluid in the most intense scenes. With quadCF, this never happened. QuadCF also considerably improved my performance in Crysis Warhead to the point where a previously unplayable setting was fairly playable.
 
That CPU must be volted up pretty high then, because even at 110V your PSU isn't much less efficient than mine.

The numbers are correct:

Each 4870x2 goes up to 320w at full load. I am not pulling numbers of of my ass:
http://www.geeks3d.com/20100226/the-real-power-consumption-of-73-graphics-cards/

Lets assume 200w for MOBO + CPU. It is a conservative guess:) 250w for really leaky setups.

Thats 840 watts

At 80% efficiency , we could easily see numbers in the 1k watts range coming out of the power socket, but a 1k watts PSU could handle the PC well enough.
 
You do realise I also owned a 4870X2 Quad setup and measured it right? I'm merely observing that GotNoRice is seeing higher power numbers than me. The PSU concerned measures 87% at high load (I'm assuming 110V due to the connector style) and mine (due to 230V) around 88%.
Typical load in games for the 4870X2 is about 290W, and in stress tests around 380W. A heavily overclocked quad + board uses about 100W at moderate load, and around 250W in a stress test.
Add these up with games and you get 680W in games before other hardware (680/0.88 = 773, close to the 830 figure I posted. Allow for HDDs and it makes sense) and 1010W with a full stress test. 1010/0.88 = 1147. Same here.
 
Back
Top