Originally Posted by Darkchild
ive honestly never ran it. what actual performance number can you get from a stress test that could possibly screw your card that some other less dangerous test cant show?
Furmark benchmarks really just stresses the maximum "potiental" performance of your card very HARD at 100% usage at a consistant pace and will report a score, max temps and FPS.
It doesn't fluctuate testing or simulate everyday in game situations that your card may encounter like 3Dmark or Heaven does.
You'll also find that the card may pass a run in Furmark but fail in 3Dmark or Heaven so it's not recommended as it lacks in-depth stability testing.I use furmark for quick runs to see if a GPU will either "hit" a clock or crash immeadiatly. I also use it to tweak GPU airflow but wouldn't constantly run benchs with it because its pointless abuse to the GPU hardware and the info obtained is of little or no benifit to my task.
Is it dangerous? I've never killed a GPU with it but stressing a GPU for long periods like one would stress a CPU with Prime95 is unecessary for the GPUs in my opinion. There are more suitable softwares to find a balance between performance & stability.
Hope this clears things up a bit