You can create your own profiling system just using timing functions like QueryPerformanceCounter(), and a log system. Here is an example of how it works:

http://www.cryer.co.uk/brian/delphi/howto_time_code.htm

OTOH, this software called my attention, but I haven't tried it:

http://www.intel.com/cd/software/pro...une/239144.htm

If someone have tried it, please write about it.