Overclocked video cards, better for CUDA

First posting here, one silly question:
Does one video card, namely 8800gt performs faster in CUDA calculations if it’s overclocked?

Thanks!

Of course individual kernels will be faster (how much depends on how you changed clocks and where the bottlenecks in your app are), but you may hit other bottlenecks and stability problems as a result.

Does some particular clock counts in speeding computations? Gpu, memory, shader?I could try to increase one of them higher if it counts.Now it’s overclocked in game style, all clocks increased in tandem.

Again thanks!

Shader and memory are the two you will care about.

I will care about them, then External Media.

Thanks Murray