Tracking NH Prof performance problem on aggressively active applicatio

I mentioned before that NH Prof has a problem with dealing with a lot of activity from the profiled application. In essence, until it catch up with what the application is doing, there isn't much that you can do with the profiler. I consider this a bug, and I decided to track that issue with some verifiable results, and not just rely on hunches. The results were... interesting. The actual problem was in the place that I expected it to be, but it was more complex than expected. Let me try to explain.