[JP_Singer]

My new rig
Vista 64bit Ultimate
Antec Case
650 Watt Crossfire ready Antec PSU
GA-EP35-ds4 775 with the latest Bios update
Q6600 G0 SCLAR overclocked @3.3 46c under load
ZALMAN CNPS9700 Cooler
2x2gig=4gig OCZ Platinum 800
Visiontek 3870x2 HD Crossfired with a Visiontek 3870 HD
500GB Maxtor 16mb hard drive
Pioneer DVD 212D Burner
24 Samsung Monitor 5ms

Ok so I built my new rig and installed all the latest drivers with ATI's 8.4s

My first Default 3dmark06 score was with only the 3870x2
3Dmark scored 16800

My second Default 3Dmark06 score was with the 3870x2 crossfired with the 3870 bridged.
Scored 18465

Yes,
Crossfire enabled in CCC with Advanced AI

I also completly removed the 8.4s and tried the 8.3 with and without hot fix showing simular results with 8.4s.

Crysys with latest patch:
Crysis Benchmark score Very High Setting Average 15fps with just the 3870x2 installed.

Crysis Benchmark score Very High Setting Average 19fps with the 3870x2 crossfired with 3870 installed.


Crossfire is enable and they do show up in device manager and in the CCC overclock util.

There is no setting in the bios for crossfire but my software shows it is enabled but the hardware does not act like it.

If anyone has any ideas why my scores suck with crossfire then please let me know.

Hopefuly I overlooked something.

THanks

[ghost_recon88]

CPU is probably bottlenecking the 2nd card. Get it up to 3.6GHz and run again. Also, I assume you mean HD3870, rather then HD6870? Because if you have a HD6870, can you sell me one? Also on that board, its a 16/4 configuration with the 16x PCI-E slots, rather then 16/16 like the X38 and X48 chipsets have. Only having 4x bandwidth available to a HD3870 will bottleneck it a bit. You do have the HD3870x2 in the 16x slot right?

[JP_Singer]

HD6870 LOL

I changed it to 3870

I do have the 3870x2 in the 16x slot.

Would one lane being 4x vs 16x in crossfire realy make that big of an impact on going from a single card to crossfire?

THanks

[ratbuddy]

Originally Posted by JP_Singer

HD6870 LOL

I changed it to 3870

I do have the 3870x2 in the 16x slot.

Would one lane being 4x vs 16x in crossfire realy make that big of an impact on going from a single card to crossfire?

THanks

Not really. Those scores are right where they should be for your setup.. Clocking the CPU a bit more would give higher scores, but remember it's just a synthetic benchmark. Gaming will not really show any difference. You'd be better off with a higher clocked dual core for gaming..

[vixro]

16/4 and your CPU speed both bottleneck quadfire easily.

[ratbuddy]

Originally Posted by vixro

16/4 and your CPU speed both bottleneck quadfire easily.

He MIGHT pick up 1k points by switching to an X38 board or something, but in real game performance, it's not gonna make enough difference to be worth it, unless his favorite game is 3dmark..

[ghost_recon88]

Its a lot more then 1k points, Tomshardware had a recent article comparing 4x, 8x, and 16x bandwidth. http://www.tomshardware.co.uk/PCI-Ex...w-29948-3.html. A 33 and 25% performance hit with the cards below. Sure with Crossfire you don't get twice the performance as you would with a single card, but the scaling is a lot better then 1k points worth.

[ratbuddy]

I'm not good with math, but the way it seems is that only 33% of his graphics firepower is in that x4 slot, so even if it shows a gain of 33%, that's only 11 percent overall.. Assuming a perfectly optimal situation, the most it'll gain from switching to x16 would be 2000 points, and I'd be surprised if it was that much. Gamewise, +-11% fps isn't going to make or break anything. I don't think going from 30 to 27 or 33 fps makes any difference (all unplayable ) and again, I doubt it's going to scale perfectly like that..

The numbers do seem to indicate it would be worth more than 1000 points though, I'll concede that point. I stand by my assertion that it doesn't matter unless all he plays is 3dmark

Another way to look at the slot bottleneck: The 3870X2 in the x16 slot did 16800, adding the 3870 to the x4 slot increased the score by 1665, for a total of 18465. Let's say those additional 1665 points are 25% lower than they would be in a x16 slot, it's gonna be 1665 * 1.33, or 2214 points. add that to 16800, and you get an expected score in trifire on a dual x16 board of 19014.. Now I'm curious to see which case is true. Oh well.

[Mr.Guvernment]

can you xfire a x2 with a single card?

Visiontek 3870x2 HD Crossfired with a Visiontek 3870 HD

[Mr.Guvernment]

Originally Posted by ratbuddy

Not really. Those scores are right where they should be for your setup.. Clocking the CPU a bit more would give higher scores, but remember it's just a synthetic benchmark. Gaming will not really show any difference. You'd be better off with a higher clocked dual core for gaming..

a 3.3ghz quad is more then sufficient for gaming. vs say a 3.6ghz dual core, the diff wouldnt be noticeble but the other all system smoothness likely would be in the quad vs the dual

[JP_Singer]

Thanks to all that have replied.

But,

If all I am going to see from my crossfire is a 2000 point increase in performance then what is the point? That is less than a 2% gain.

So I would be better off selling or trading with someone the q6600 and getting a e8400 and overclocking it to 4.0ghz.

I am starting to think I should take the 3870 back and stay with just the x2

[Mr.Guvernment]

remeber 3dmark is not a game, play a REAL game see if you notice any diff. dont base it off a synthetic benchmark.

look for reviews of Xfire on games you play, see if it offfers what you want, which technically you should of done before buying into xfire.

[JP_Singer]

Well thanks but,

3dmark06 does a good job as far as testing the performance of your video card and I do understand your point.

I saw just about as much increase with 3dmark06 as I did with Crysis.

I was disapointed to see my top frames with very high setting in crysis were 19fps.

I did my research and saw other people getting 35fps to 55fps with a set up like mine But they may have been way overclocked.

I guess I need to start paying more attention.

[vixro]

you were trying to play crysis on very high. No card setup can run very high yet and still be playable. Try high instead.

[JP_Singer]

THe only way I can play crysis is with everything very low.

Something is wrong.

[ratbuddy]

Originally Posted by vixro

you were trying to play crysis on very high. No card setup can run very high yet and still be playable. Try high instead.

Mine can, but then again, my primary display is only 1280x1024

[Mr.Guvernment]

most people who are getting good FPS in crysis did the ini tweaks that makes high look like ultra high with out the cost of the FPS.

[Mathis]

Your scores seem a lot like mine. I too have a trifire setup. I used to have an e8400 and I was scoring almost the same as what I am with a q9450. The only difference is my SM2.0 and SM3.0 scores are lower and my CPU score is significantly higher. I gain approximately 1600 3dmarks from the extra 3870, but my games have much better performance gains, especially Crysis. The 3dmark scores almost had me doubtful to the point where I wanted to sell my extra card too, but try out some games and forget about 3dmark for a little while.