Using the kernel-debug package really affects the disk performance. With the Threaded I/O Tester for carrying out 64MB writes at 32 threads, the latency had jumped quite a bit when having to deal with the kernel debugging code. Disabling SELinux and Audit led just to a tiny performance improvement in this test.
Disabling SELinux and Audit led to a better advantage when it came to disk reads, though the kernel debugging support led to a smaller loss over the stock F11 kernel.
To no surprise, the best performance with PostMark was found when using the stock kernel and no SELinux/ Audit. There was a sizable gap between the stock kernel and kernel-debug.