Thermal testing was conducted by monitoring the various GPU temperatures with HWMonitor 1.14 64-Bit. The idle condition was established by recording the temperature after letting the system sit for 30 minutes with only HWMonitor launched. The load condition was established by running 3DMark Vantage two times back-to-back, followed by two back-to-back sessions of the Resident Evil 5 benchmark. HWMonitor is capable of logging current, minimum, and maximum temperature, so once all testing was completed, the maximum temperature was recorded. All testing was completed on an open testbed with relatively still air in a room with an ambient temperature of 20C.
HWMonitor is able to monitor both cards separately when in CrossFire, and it was interesting to see that the GPU sandwiched between the two PCBs didn't get any warmer than the one that should have had more airflow. That said, whether the HD5850 was used alone or in CrossFire, it isn't exactly cool at idle, but it never got very hot, either.
Power Consumption Testing:
Power consumption testing involved the use of an AC power monitor (a Seasonic Power Angel) to keep track of the AC Wattage drawn by the test system. Since all hardware other than the graphics card remained the same from test to test, any change in the Wattage can be attributed to the demand of the graphics card. Idle conditions were established by letting the system sit at the Windows desktop until a stable reading was achieved, and for load testing 3DMark Vantage was executed two times back-to-back. Peak values were recorded from the Power Angel's display and recorded in the chart below.
At idle the HD5850 is able to throttle back and consume electricity rather gently. But when it is time to perform, especially in CrossFire, you can really get the meter spinning. On a positive note, the HD5850 outperforms the previous generation HD4890, and can do so while consuming less electricity. Efficiency improvements are always welcome.