Testing (continued):
Thermal Testing:
The next test involved monitoring the GPU temperature while the cards were at idle and under a heavy load.
HWMonitor 1.08 was used to monitor GPU temperatures with the cards running on a test platform (no case) in a room with a constant 21C air temperature. The system was allowed to idle at the Windows desktop for at least one hour before the idle temperature was recorded, and then 3DMark06 was executed so that it conducted each test three times before exiting. At this point the maximum temperature reported by HWMonitor was recorded.
Much to my surprise, the somewhat generic Visiontek HD3850 wound up being the hottest card. The HD3870 TOP idled at the highest temperature, but the thermally controlled fan kept it from getting too hot under load. As I mentioned earlier in the review, the fan on the HD3870 ramps up more often than other cards I have used, but at least the extra activity/noise serves a purpose.
Power Consumption Testing:
The final test was to analyze the power consumed by each video card. The methodology involved connecting a
Seasonic Power Angel to the test system in order to monitor the Wattage drawn by the system from the power outlet. The system was allowed to idle at the Windows desktop for several minutes in order for the reading on the Power Angel to stabilize, and then the idle Wattage value was recorded. 3DMark06 was then executed once again and the Power Angel was monitored in order to identify the maximum value achieved.
A twist to my approach is that the initial test was conducted with a 64MB Radeon X300SE installed into the system specified in the configuration section of the review. This extremely low end card is about as close to onboard graphics as you can get, and would draw the least amount of power possible while still letting me see something on the screen. The point of this was to record a power consumption differential value for the three cards being compared, so we are only looking at the Wattage increase and not total system power. For reference, with the X300SE installed the system drew a total of 69 Watts at idle and 115 Watts maximum.
Looking at the chart above details that using the HD3870 TOP resulted in a 21 Watt increase in power consumption while idle, and a 98 Watt increase under a load. Under a heavy load the HD3870 requires 23 Watts more than the HD3850 TOP and 33 Watts more than the HD3850. Not much of a difference, and if your power supply can handle an HD3850 it should be able to handle an HD3870 without issue.
