(2) Sapphire 4850 X2 in Quad-fire review
Here is my bitter sweet quad-fire setup using (2) 4850 X2’s. I say bitter sweet b/c I brought my second X2 used and it has some semi-damaged memory on gpu-0 side of the card. I will only say this is not a Sapphire manufacturing error, and leave it at that. Testing was very difficult, taking a few days and I’ve already moved back to tri-fire. That being said, on to the info.
(2) 4850 X2
Installed
Installed (close up)
Catalyst Control Center (ccc): crossfireX configuration
Catalyst Control Center (ccc): Graphics hardware info
Primary Adapter
Graphics Card Manufacturer Powered by ATI
Graphics Chipset ATI Radeon HD 4850 X2
Device ID 9443
Vendor 1002
Subsystem ID E870
Subsystem Vendor ID 174B
Graphics Bus Capability PCI Express 2.0
Maximum Bus Setting PCI Express 2.0 x8
BIOS Version 011.011.000.000
BIOS Part Number 11X-B50102-X06
BIOS Date 9/16/2008
Memory Size 1024 MB
Memory Type GDDR3
Core Clock in MHz 625 MHz
Memory Clock in MHz 993 MHz
Total Memory Bandwidth in GByte/s 63.6 GByte/s
Linked Adapter
Graphics Card Manufacturer Powered by ATI
Graphics Chipset ATI Radeon HD 4850 X2
Device ID 9443
Vendor 1002
Subsystem ID E870
Subsystem Vendor ID 174B
Graphics Bus Capability PCI Express 2.0
Maximum Bus Setting PCI Express 2.0 x8
BIOS Version 011.011.000.001
BIOS Part Number 11X-B50102-X06
BIOS Date 10/23/2008
Memory Size 1024 MB
Memory Type GDDR3
Core Clock in MHz 625 MHz
Memory Clock in MHz 993 MHz
Total Memory Bandwidth in GByte/s 63.6 GByte/s
Linked Adapter
Graphics Card Manufacturer Powered by ATI
Graphics Chipset ATI Radeon HD 4850 X2
Device ID 9443
Vendor 1002
Subsystem ID E870
Subsystem Vendor ID 174B
Graphics Bus Capability PCI Express 2.0
Maximum Bus Setting PCI Express 2.0 x8
BIOS Version 011.011.000.001
BIOS Part Number 11X-B50102-X06
BIOS Date 10/23/2008
Memory Size 1024 MB
Memory Type GDDR3
Core Clock in MHz 625 MHz
Memory Clock in MHz 993 MHz
Total Memory Bandwidth in GByte/s 63.6 GByte/s
Linked Adapter
Graphics Card Manufacturer Powered by ATI
Graphics Chipset ATI Radeon HD 4850 X2
Device ID 9443
Vendor 1002
Subsystem ID E870
Subsystem Vendor ID 174B
Graphics Bus Capability PCI Express 2.0
Maximum Bus Setting PCI Express 2.0 x8
BIOS Version 011.011.000.000
BIOS Part Number 11X-B50102-X06
BIOS Date 9/16/2008
Memory Size 1024 MB
Memory Type GDDR3
Core Clock in MHz 625 MHz
Memory Clock in MHz 993 MHz
Total Memory Bandwidth in GByte/s 63.6 GByte/s
GPU-Z technical info
GPU-Z clock, thermo, and fan speed info
Testing
3Dmark 2006
3Dmark Vantage
Call of Duty 4: MW (could not test)
Call of Juarez
Crysis
Devil May Cry 4
Furmark
Left 4 Dead (could not test)
Sanctuary
Tropics
X3 Terran Conflict
3Dmark 2006:
4850 x2 = 17847 (3.8G used)
4850 x3 = 19934
4850 x4 = 19818 (116 point, 0.5% loss from 4850 x3)
3Dmark Vantage:
4850 x2 = P10537 (3.8G used)
4850 x3 = P13251
4850 x4 = P13784 (533 point, 4.7% gain over 4850 x3)
Call of Juarez:
1680 x 1050, highest quality settings
4850 x2 = 61.4 average fps
4850 x3 = 92.7 average fps (31.3 fps, 50.9% gain over 4850 x2)
4850 x4 = 102.7 average fps (10 fps, 10.7% gain over 4850 x3)
Crysis:
1600x1200
4850 x2 = 33.9 average fps
4850 x3 = 41.9 average fps (8 fps, 23.5% gain over 4850 x2)
4850 x4 = 41.6 average fps (0.3 fps, 0.6% loss over 4850 x3)
1900x1200
4850 x2 = 29.2 average fps
4850 x3 = 35.3 average fps (5.7 fps, 20.8% gain over 4850 x2)
4850 x4 = 37.1 average fps (1.8 fps, 5.0% gain over 4850 x3)
2560x1600
4850 x2 = 17.8 average fps
4850 x3 = 23.5 average fps (6.1 fps, 32.0% gain over 4850 x2)
4850 x4 = 26.1 average fps (2.6 fps, 11.0% gain over 4850 x3)
Devil May Cry 4:
1680x1050
4850 x2 = 170.2 average fps
4850 x3 = 242.5 average fps (72.3 fps, 42.2% gain over 4850 x2)
4850 x4 = 268.0 average fps (25.5 fps, 10.5% gain over 4850 x3)
Furmark:
1440x900
4850 x2 = 197 average fps
4850 x3 = 284 average fps (87 fps, 44.1% gain over 4850 x2)
4850 x4 = 287 average fps (3 fps, 1.0% gain over 4850 x3)
1680x1050
4850 x2 = 152 average fps
4850 x3 = 219 average fps (67 fps, 44.0% gain over 4850 x2)
4850 x4 = 232 average fps (25.5 fps, 10.5% gain over 4850 x3)
Sanctuary:
1680x1050
4850 x2 = 111.3 average fps
4850 x3 = 167.4 average fps (56.1 fps, 50.4% gain over 4850 x2)
4850 x4 = 194.4 average fps (27.0 fps, 16.1% gain over 4850 x3)
Tropics:
1680x1050, DX9
4850 x2 = 23.9 average fps
4850 x3 = 80.0 average fps (56.1 fps, 334.7% gain)
4850 x4 = 106.2 average fps (26.2 fps, 32.7% gain over 4850 x3)
X3 Terran Conflict:
16800x1050
4850 x2 = 94.7 average fps
4850 x3 = 97.7 average fps (3 fps, 3.1% gain over 4850 x2)
4850 x4 = 96.5 average fps (1.2 fps, 1.2% loss over 4850 x3)
Conclusion: So is quad-fire worth it? Absolutely not. Well at least not at this moment in time. There are few issues that are not quite ready for quad-fire. The one most thought of by people, 8x/8x pci-e, is not one of them. From what I can tell, the problem issue with quad-fire is half in part of what ati has told us for a year and what they are still working on. The 100% use of the 4th gpu. Today, even with the current catalyst drivers, the use of 4th gpu is not used in many applications. In some of the tests I’ve done listed above, I’ve noticed in gpu-z, not only 1 but sometimes 2 gpus were not fully active. Ok to sum up this report, if you have an X2 or just have an HD48xx card and are looking for more power and the best band for the buck; go for tri-fire. Either get another HD48xx to match your X2 or get and X2 to match you single HD48xx. Quad-fire does work and has a small and I mean very small advantage over tri-fire, but its just not worth the extra money, power consumption and system heat. Hope this review gives everyone a little insight on multi-ati goodness; this has been fun.
ATI for life.
