HD4870 Crossfire = No performance change?

[humous]

I have purchased 2 radeon HD4870 and have them enabled with, what I can only presume is all the right bits. Crossfire-enabled MB, crossfire links, etc.

However, I can't see any difference in Game speeds at all. I have tried some stuff with one 4870 card in; Crysis, GTA4, Total war, 3dMark; and the system performs ok. But when I put in the second 4870 card, it makes no difference at all. Maybe a 1% increase in 3dMark.

Is this normal? Or is there something I'm not doing right. Bit gutted really as it seems to have been a huge waste of money.

I'm currently running:
P45 Platinum MB
Intel Quad-core 2.33Mhz
2Gb DDR2 ram
2 x HD4870 1GB
ATI Catalyst 9.3 (the latest one)

Any help please!

 

[Maziar]

Something is wrong here,
what resolution do u play at ? also make sure u have enabled CrossFire.
Download GPU-Z and check with it too.

 

[humous]

Thanks for your reply!

I play at 1080p mostly.

Crossfire is enabled in the ATI Catalyst. Catalyst A.I. is enabled (apparently)
3dMArk score is 13100
2.0 4800
3.0 6400
CPU 3400
roughly - can't remember them exactly

Is there some kind of test to see the difference for 1 card vs Crossfire. I'm sure it's just not being enabled for some reason.

GPU-Z shows the 2 cards, saying for AtI crossfire:
"Enabled(2 GPUs)(unsure on Vista64)"

 

[humous]

Update:
I ran 3dMark with GPU-Z running and displaying GPU load for both cards.

3DMark Score 12749
SM 2.0 Score 4775
SM 3.0 Score 6836
CPU Score 3019

GPU load for each card seemed to be the same, indicating they're both working in parallel correctly - so why no performance increase?

 

[Maziar]

Test the cards one by one and check if they both give u the same performance in single mode.

 

[humous]

Thanks for your reply.

Yes, they do...

 

[wh3resmycar]

Crysis, GTA4, Total war,

do these games actually gets perf increase from xfire? i thought crysis was a xfire bust (but crysis:warhead isnt).

 

[rags_20]

do these games actually gets perf increase from xfire? i thought crysis was a xfire bust (but crysis:warhead isnt).

Even if a game is a CF bust, it will show at least 10-15% performance increase (those games)

forget the games, look at 3DMark, it sure uses CF.

Anyway, I'd say there could be some prob. with the link width. Not too sure, but could this be due to running them a x8 - x8?

 

[humous]

Thanks for your replies...

I think I just want to know what games do well on Crossfire. If you say that Crysis warhead performs well with Crossfire, I will try it with Single vs. Crossfire configurations and see the difference.

I wasn't expecting everything to be twice as fast (or even the 2.7x the marketing seems to suggest) but I do want to see some difference. Just to make sure that something is actually happening.

Are there any other games that work especially well with Crossfire?

 

[humous]

Also, there seems to be lots of people complaining about crossfire in roughly the same way as me.

Are there any crossfire users out there experiencing good performance? I'm not rubbishing crossfire or anything - I just want to know if I should keep trying to get it to work or just be happy with the 1% performance increase...

 

[rags_20]

Forget the games. You will not get a performance increase in games if you don't in 3D Mark. It takes full advantage of CF. Anyway, there was a thread a while ago about a guy benchmarking 4870 3CF. In certain games, there were performance increases. That might help.

 

[humous]

Here's an odd thing...

I get the same 3DMark score and framerate when running at 1080p as I do when running at 640x480.

Surely, this isn't normal? Maybe the same score, but there shouldn't there be a much better framerate?

 

[daedalus685]

Usa vantage if you can. 06 tends to be severely limitted by the cpu which is why you see teh same performacne at all resolutions. (you might consider overclocking it)

With vantage and CF enabled I see roughly 60% improvements on P setting. (I cant remember the numbers exactly as I dont have it installed currently.. but it was surely clear.

 

[humous]

Thanks everyone for all your help. It's all academic now, I'm afraid as my Motherboard has blown, it seems. Just makes three low pitch beeps, then turns off, then tries to turn on again. And repeat...

I'm going to return it as I'm not sure now if this motherboard (p45 platinum) is fully compatible with the crossfire setup. - even though it says so on the box. It mentions something about a whopping 8% VGA performance increase, which seems to be about the performance increase I'm actually getting.

So, I will be on again to let you know.

Any ideas for a Intel Core 2 Quad - compatible crossfire motherboard?

Thanks for all your help, guys...

 

[rags_20]

Yeah, Maximus Formula should do it.
Are you sure you connected the CF cable properly and enabled CF in the Control Center?

 

[RazberyBandit]

Here's an odd thing...

I get the same 3DMark score and framerate when running at 1080p as I do when running at 640x480.

Surely, this isn't normal? Maybe the same score, but there shouldn't there be a much better framerate?

The downloadable, free version of 3DMark's software always runs at the same resolution, and you can't change that. Even if you set your monitor to a specific resolution, the software runs the tests at whatever pre-set resolution is hard-coded within it. So the answer to your question is, yes, this is normal. Just a little something most users overlook when it comes to benchmarking with a free synthetic like 3DMark's products. Only the PAID versions of their software allow things like resolution to be altered.

Sorry to hear the mobo died... an X38 or X48 is probably the route to go. You'll likely see prices in the $150 to $200+ range.

 

[Kari]

lets hope it was the bad mobo all along...

crossfirex should scale pretty well with modern drivers (at least if you have the cpu-power to back it up)
http://www.xbitlabs.com/articles/video/display/crossfirex-sli-core-i7_10.html#sect0