Compute Visual Profiler not working

I have written a CUDA program and want to profile it using Compute Visual Profiler which is installed when installing CUDA toolkit 3.1.

The problem is that the profiler seems to always timeout. It does not matter whatever I set as a timeout. It always seems to take longer time than specified “max execution time” in session settings.

I always get this message in output window:

Program run #1 was aborted after maximum program execution time duration of 30 seconds.

Does anyone know how to fix this?

My config:

Geforce GTX 470 (fermi)
Windows 7 x64
CUDA toolkit 3.1
display drivers: 297.21

EDIT: If I let the profiler run til it finishes I get these error meeages:

There is some error in profiling. The profiler output file is empty.
Error in reading profiler output.

Edit 2: Are there anyone with a fermi card that can use Compute Visual Profiler to profile?

Edit 3: I got it to work. The problem was that I had a getchar() call in the end of my main function. The profiler just waits for a keyboard input but none is coming so it waits until timeout is reached

Increase the execution time limit in the session settings to something larger.

Increase the execution time limit in the session settings to something larger.

I have tried larger values. tried 1000. The profiler just runs extremely slow

I have tried larger values. tried 1000. The profiler just runs extremely slow