I have experienced the problem that the SystemAnalyzer would not always load all threads. I now found out by coincidence that this happens if not enough time has passed between thread creation and Kernel initialization. At least that is what I am assuming.
I created a for loop counting to 100000000 after the thread creation and now the are loaded every time in the SystemAnalyzer. I would really like to understand why this happens. This is probably not the best way to make sure that it works. I am actually initializing a lot in between and I am creating the threads right after the initialization of the Kernel.