AMD Radeon HD 7950 3GB CrossFire ReviewFri, Feb 10, 2012 - 12:00 AM
For testing power consumption, we took our test system and plugged it
into a Kill-A-Watt power meter. For idle numbers, we allowed the system
to idle on the desktop for 15 minutes and took the reading. For load
numbers we measured the peak wattage used by the system while running
the OpenGL benchmark FurMark 1.9.2 at 640×480 resolution. We also ran the game HAWX 2 and looped the benchmark three times and recorded the highest Wattage seen on the meter.
When trying to get load temps and load power consumption readings we
tried to use Furmark, but ran into some problems with both Furmark v1.9.1 and v1.9.2. With CrossFire
enabled both cards would run when Furmark was launched, but the
secondary card wasn’t running at full speed and didn’t appear to be
under load. We contacted AMD about this issue and the didn’t have an
answer for us.
Power Consumption Results: Our test system with a pair of AMD Radeon HD 7950 video cards running in CrossFire was observed using ~450 peak Watts at the wall during gaming. Idle performance of the cards was impressive and on par with what we saw with a pair of Radeon HD 7970 cards in CrossFire with an idle right around 100 Watts.
The AMD Radeon HD 7900 series has a new technology called ZeroCore Power, which shuts down the GPU during periods of long idle (when the screen goes to sleep). When the GPU goes to the power state the fan stops spinning and the GPU core itself consumes 0W while the rest of the PC is still running. This is a really interesting feature and we found that it works great. After the system is at long idle the video card basically shuts down and goes to sleep. With a single AMD Radeon HD 7950 reference card in our test system the idle power was down to 86 Watts and with two cards in the system it was 92 Watts.
One of the main reasons AMD developed this technology was from feedback from CrossFire users as they didn’t need all those cards running when at idle as they produce noise and heat when it isn’t necessary.