Sign in with
Sign up | Sign in

Cross-Check

Comparison: Factory-Overclocked Versus Reference Graphics
By

We decided to run a few additional tests with several other, factory-overclocked graphics cards because we wanted to be sure that the results we’ve been seeing on MSI’s R5870 Lightning TwinFrozr II would apply to other boards with aftermarket enhancements as well. The two other cards also reduce clock speeds at idle, so long as they operate at their preset clock speed settings--we can be sure because the GPU temperatures stay below 45°C.

A quick look at the temperatures under 3D load make clear which cards are based on AMD’s Radeon HD 5870 reference design. All cards that reach 75°C or more are based on the same layout and cooling. The dual-fan solutions are in between 63 and 66°C.

The higher you overclock the graphics card, the more power will be required, even in 2D, as the cards don’t throttle clock speeds all the way down to where they would end at factory clock speed settings.

The 3D power consumption results also show very obviously which cards were factory-optimized for higher clock speeds. Most factory-overclocked cards run at slightly increased GPU voltage to maximize reliability, which adds an additional ~20 W to the total power consumption on these systems.

You don’t need fast fan speeds if the graphics card throttles clock speeds at idle. Card vendors that do the best job optimizing fan speed stay below 36 dB(A).

3D peak loads reveal the differences between reference cooling and the optimized products: it’s high air flow and noise versus optimized power and low noise. In an ideal case, overclocked graphics cards with optimized cooling solutions aren’t any louder than reference cards at reference speeds. If a factory-overclocked card goes for fast fan speeds right away, then you can be sure that it prepared the product for serious overclocking.

React To This Article