Hi,
I have tried two different Crossfire setups, using the same rig, but both of them performed so badly I started a thread on the AMD Game Forum - then someone locked my thread!?
http://forums.amd.com/game/messageview.cfm?catid=262&threadid=139433&enterthread=y
I would really appreciate a bit of feedback from anyone else who has tried 3-way crossfire - setups such as HD5970 + HD5870 don't count, neither do HD4870x2 + HD4870, as they are technically 2-way setups.
I'm talking 3 GPUs, 3 PCBs and 3 PCIe slots.
My test rig was a Foxconn ELA P45 (3 PCIe @ x8 speeds), I used 4GB RAM @ 1000MHz and a E8400 @ 3.8GHz. (900W PSU)
The cards I used were;
3x HD3870s (GDDR4)
3x HD4890s
Both setups destroyed 3DMark06 & Vantage, both sucked when it game to games.
What I found was 2-way Xfire worked great, but when I enabled the 3rd card in either setup, the frame rate became crazy erratic - with a max frame rate higher than 2-way, but the min dropped to lower than a single card!?
Averaging at no better than a single card.
I found an old Custom PC mag in which they compared 3 and 4 way Xfire v SLI, and my results with the HD3870s were identical to the magazines own results. With anything more than 2-way becoming a hinderance not a help.
2-way Crossfire seems to either work or not, it never gets WORSE than a single card.
So I've tested the best HD3000 and HD4000 series cards, which should have the better drivers - being well seasoned cards.
I know that it may be game specific but I hardly see the benefit of including a 3rd card that I am rarely going to use.
Anyone found a 3-way setup that works? Is it the ATI drivers or the games?
I'm guessing that it's the games - and not me...
I have tried two different Crossfire setups, using the same rig, but both of them performed so badly I started a thread on the AMD Game Forum - then someone locked my thread!?
http://forums.amd.com/game/messageview.cfm?catid=262&threadid=139433&enterthread=y
I would really appreciate a bit of feedback from anyone else who has tried 3-way crossfire - setups such as HD5970 + HD5870 don't count, neither do HD4870x2 + HD4870, as they are technically 2-way setups.
I'm talking 3 GPUs, 3 PCBs and 3 PCIe slots.
My test rig was a Foxconn ELA P45 (3 PCIe @ x8 speeds), I used 4GB RAM @ 1000MHz and a E8400 @ 3.8GHz. (900W PSU)
The cards I used were;
3x HD3870s (GDDR4)
3x HD4890s
Both setups destroyed 3DMark06 & Vantage, both sucked when it game to games.
What I found was 2-way Xfire worked great, but when I enabled the 3rd card in either setup, the frame rate became crazy erratic - with a max frame rate higher than 2-way, but the min dropped to lower than a single card!?
Averaging at no better than a single card.
I found an old Custom PC mag in which they compared 3 and 4 way Xfire v SLI, and my results with the HD3870s were identical to the magazines own results. With anything more than 2-way becoming a hinderance not a help.
2-way Crossfire seems to either work or not, it never gets WORSE than a single card.
So I've tested the best HD3000 and HD4000 series cards, which should have the better drivers - being well seasoned cards.
I know that it may be game specific but I hardly see the benefit of including a 3rd card that I am rarely going to use.
Anyone found a 3-way setup that works? Is it the ATI drivers or the games?
I'm guessing that it's the games - and not me...