Power Consumption



The system power consumption was measured using a Brennenstuhl PM 230 electrical energy meter. In real life the power consumption is not measured, it is being calculated by measuring the AC voltage at the back of your pc's power supply multiplied with the measured current flowing through the mains cable and multiplied with the power factor which occurs when using capacitive or inductive loads on alternating current (AC). In any case, our device is no professional equipment, our results can be off be 5%, if not more, but at least we are not left guessing the power consumption. Do understand that this is the total system powered measured at the back of our PC's power supply, it is not an indication of what to expect from a single computer component. Here is what we got for our tested products:



The slightly higher clock speeds and few extra components make the 9800GTX come out just a bit worse than the 8800GTS 512MB, in 'idle' status we could not measure any difference between the two. At a total system power consumption of 360 Watt the 9800GTX requires a lot more power then the midrange 9600GT, just don't forget that the GTX comes with double the amount of shaders and a lot more performance.

Temperature



Rivatuner is the tool every video card geek should have in house, it not only allows adjusting clock rates to higher levels, it also gives us the opportunity to log the GPU temperature over longer periods of time. We used Rivatuner to log the GPU temperature while 3D Mark 2005 in loop was used to stress the video card, here are our results:



Even though the 9800GTX consumes the most power, it comes out as rather good in the temperature chart. There is lots of cooling headroom left before the GPU will ever be overheated, the heatsink can take quite some more heat and if that isn’t enough there is always the auto-fan control which will increase the fan speed whenever the GPU temperature is getting closer to critical levels.

Noise



Noise created by fans and spinning hard disks is not what everyone cares about, though some people are really keen on dead silent pc's, therefore we added a noise chart. Noise is measured with a Smart Sensor AR824 digital sound sensor 50cm away from our Antec PC housing. We stopped all fans in order to give an exact representation of the noise that is being created solely by the tested video card.



After longer test runs I went back to have a look at the Rivatuner temperature log window and I noticed how the GPU temperature dropped rather fast. Although the fan speed in Rivatuner remained at 100% constantly, I'm quite sure that the fan did change its speed. Measuring the sound level cleared up any doubts, the fan was indeed auto controlled. Trying force fanspeed in Rivatuner didn't work out that well, I went on using Galaxy's Xtreme Tuner but even there the fan speed started to have a life of its own after few minutes testing. I think we have to wait for newer versions of our software with support for the new controller IC, if we want to regulate fan speed ourselves that is.

Out-of-the-box the card never was noisy, it was noticeably during heavy 3D load, but with high-end CPU's, high-end mainboards and other high power components around the noise level created by the video card only might not even be noticed. In the end, we’re talking about high-end, where performance comes over noise and power consumption for most people.