Viewing page 7 of 8 pages.
Previous 1 2 3 4 5 6 7 8 Next
Test Results: 2 x ATI Radeon HD 4870 (CrossFireX)
2 x HD 4870 CrossFireX test system.
|
|
System Measurements (2 x HD 4870)
|
|
System State
|
Idle
|
CPU + GPU Load
|
|
System Fan Speeds
|
7V
|
9V
|
12V
|
|
CPU Temp
|
33°C
|
47°C
|
46°C
|
42°C
|
|
SB Temp
|
54°C
|
64°C
|
63°C
|
62°C
|
|
HD Temp
|
29°C
|
28°C
|
28°C
|
27°C
|
|
GPU #1 Temp
|
77°C
|
89°C
|
89°C
|
89°C
|
|
GPU #1 Fan
Speed
|
990 RPM
|
2300 RPM
|
2300 RPM
|
2300 RPM
|
|
GPU #2 Temp
|
64°C
|
84°C
|
83°C
|
83°C
|
|
GPU #2 Speed
|
1000 RPM
|
1820 RPM
|
1810 RPM
|
1800 RPM
|
|
SPL@1m
|
24 dBA
|
33 dBA
|
33 dBA
|
34~35 dBA
|
|
System Power
|
188W
|
508W
|
508W
|
507W
|
|
CPU fan set to 100% speed.
Ambient temperature: 22°C.
|
Despite the addition of a second HD 4870, the noise level at idle of the CrossFireX system was just half a decibel higher. There was only one significant change in temperature, the Southbridge, which warmed up by 8°C as a result of its proximity to the GPUs.
On load we found that the sweet spot for the fans was 9V with the system emitting 33 dBA@1m. At this setting, it didn't produce any more measurable noise than 7V, but less than 12V. The thermal difference between these three speeds had almost no effect on the GPU though; the temperature and fan speed of the video card remained mostly unchanged.
Our HD 4870 CrossFireX test system measured 24 dBA@1m when idle and 33 dBA@1m on load.
|
The acoustics of the CrossFireX system were close to that of the single HD 4870 configuration, but higher in volume obviously. The jump from 26~27 dBA@1m to 33 dBA@1m was significant, but the noise character was similar, just higher in pitch. While not entirely unpleasant, headphones would be advised.
|
CrossFireX Configuration Comparison (Load)
|
|
Case
|
In Win Maelstrom*
|
Zalman Z9 Plus*
|
In Win BUC
|
LanCool PC-K59
|
|
Fans Speeds
|
top, rear & side @9V
|
top & rear @9V, front @7V
|
top, rear, front @9V
|
top, rear, front @9V
|
|
SPL@1m
|
30 dBA
|
30~31 dBA
|
32~33 dBA
|
33 dBA
|
|
CPU Temp
|
43°C
|
47°C
|
46°C
|
46°C
|
|
SB Temp
|
51°C
|
61°C
|
63°C
|
63°C
|
|
HD Temp
|
35°C
|
33°C
|
30°C
|
28°C
|
|
GPU #1 Temp
|
85°C
|
86°C
|
89°C
|
89°C
|
|
GPU #1 Fan
Speed
|
1720 RPM
|
1780 RPM
|
2050 RPM
|
2300 RPM
|
|
GPU #2 Temp
|
82°C
|
83°C
|
85°C
|
83°C
|
|
GPU #2 Fan
Speed
|
1620 RPM
|
1590 RPM
|
1780 RPM
|
1810 RPM
|
|
SPL@1m
|
30 dBA
|
30~31 dBA
|
32~33 dBA
|
33 dBA
|
|
CPU fan set to 100% speed.
All temperature results adjusted to 22°C ambient.
|
*Note: After analyzing the data we've collected over the last year or so, we've found that in the CrossFireX configuration, our HD 4870 cards have been putting out more heat and requiring higher fan speeds for the past couple of months. We speculate that this may have been caused by a driver update or the fan performance degrading over time (we push them very hard for long stretches of time in less than optimal conditions).
Whatever the reason, the HD 4870 coolers are now spinning 200~300 RPM faster at minimum to maintain the same temperature in recently reviewed cases, a much larger difference that we would expect considering the subjective level of airflow provided by said cases. As a result, older cases like the Maelstrom and Z9 Plus have lower noise measurements by 2~3 dB and thus we don't believe they should be used for direct comparison. We've greyed them out in the comparison table above to make that clear.
Our CrossFireX configuration in the PC-K59 ended up being comparable to the In Win BUC in thermal performance, but at the cost of a little extra noise. In the K59, the stock cooler of the HD 4870 in the upper PCI-E 16x slot ended up spinning 250 RPM faster to keep the temperature the same. It is likely that the BUC's large side panel vent gave it a cooling advantage, enough that it outweighed the level of extra noise escaping. This is interesting as the opposite was true of the same test system with just a single HD 4870.
| Help support this site, buy from one of our affiliate retailers! |
|
