On one side we have the theoretical specifications of the manufacturers (TDP and Maximum Board Power), which mostly are valid for the reference design. They depend on the graphics card and therefore don't take the power dissipation of the power supply unit into account.

On the other side we have our measurements that apply to a full PC and therefore contain the power dissipation of the power supply unit. So PCGH records the amount of energy the PC requires and what's on the electricity bill at the end of the month. In our test PC we use an economical AMD processor, an Athlon X2 BE-2350, so that the differences between the graphics cards are visually noticeable and are not overlaid by a high basic load.

As "benchmarks” we use 3DMark 03 and Furmark, to simulate the extreme case for the graphics cards. Sometimes the results differ quite a lot and that is an indication that a single test possible would not be appropriate for simulating the load of 3D games. In Furmark for example the HD 4870 surges ahead, while in 3DMark 03 it takes position below a Geforce 8800 GTX. In 3DMArk 03 on the other hand the 9800 GX2 is leading the field while it is on the fifth place in Furmark only - here only one GPU is working.

If you ask yourself why the HD 4870 X2 isn't on the list: Although we have some theoretical values from AMD, we unfortunately don't have results that are comparable to those of the other cards. We hope to deliver those results in a future article.