Shiny is a lightning fast, fully documented & by-far-easiest-to-use C/C++/Lua profiler with no extensive surgery. Results are smoothed & shown in run-time as a call-tree or sorted-by-time. Output also renderable as graphs in Ogre3D or your custom engine
At an early stage of development. Application/middleware usage monitoring/reporting. Focus on business transactions/use cases to help communicate with users. Small footprint, easy to use GUI and a level of statistical analysis to summarise the data.