Power, Heat & Noise

Here' we'll take a look at a few of the environmental effects the graphics board has on its surroundings and the rest of the system by looking at the Power usage, Heat and Noise outputs. Due to the measuring equipment we'll not be able to give definitive quantities solely pertaining to the graphics board, as we are unable to entirely isolate it, so each of the metrics are taken at a system level, however where the same system is used and only the graphics board swapped out we can begin to gauge the effects the boards have in relation to one another.

Note that due to the differences in system used, and the different slot locations of the graphics boards in the PCI Express and AGP systems, we shall only look at the results from the PCI Express system here.

First up we'll take a look at power output. In this case we are using the system in an enclosed midi-desktop case and measuring the system power usage at the main in both idle (not a desktop activity) and 3D scenarios. For the 3D use we are running Game tests 2, 3 & 4 from 3DMark03 as these have a fairly low CPU, but high graphics utilisation - the tests were run through once and repeated with the measurement taken on the second run on Game Test 2.

Looking at the power usage we see that whilst idling, rendering the 2D desktop, the system power usage isn't that different between the 6200 and the 6600 GT, standing within the 120-130W range. Under a high rendering usage scenario the system power draw only increases by 23W with the 6200, not too surprising seeing at this is a fairly low end board, whereas with the 6600 GT the system power draw is nearly doubled under load - bear in mind that the rendering performance on the 6200 means that there will be a lot more free CPU cycles, basically waiting for the 6200 board to render a frame, hence the system power draw related to the CPU and other systems will be less, however this is minimised by the use of 3DMark03 which places the rendering onus firmly on the graphics processor.

When running the power utilisation tests we also ran the case heat test at the same time taking the idle temperature a little while after the system was booted, just running the desktop, and the high utilisation temperature when 3DMark03's Game Test 2 ran. The thermal probe was placed at the back of the case, a little above the graphics card slot and was in the same place for each board tested.

Looking at the temperatures we see that the increase in temperature from idle to load situation is one degree less than that of the 6600 GT, however the 6200 is starting from a lower system temperature anyway - with portions of the chip disabled and a lower core speed its not generating as much heat as the 6600 GT is.

Finally we measured the case noise and took a reading when idle and with high utilisation under 3DMark03 rendering. In this instance the Sound Level Meter was placed at the bottom of the case - this is not indicative of actual fan noise from the graphics board, but the whole system noise from within the case; while we can't gauge the actual noise of the graphics board fans we can test to see what effects there are on the overall noise produce between each board and what differences there are between high and low graphics utilisation.

Due to 6200's lower performance, hence less power draw and heat output, in relation to the 6600 GT, the cooling solution can likewise be scaled down. Here we see that in both idle and load situations the 6200 is quieter than the 6600 GT. Due to the 6200 being aimed at the low end markets, where cost is a significant concern, it would appear that the temperature sensing circuitry is not utilised on the 6200 board as the noise doesn't change between idle and load situations, indicating that the fan is spinning at the same speed, unlike 6600 GT which varies the fan speed dependant on the core temperatures it is running at at any point in time.