I have the proprietary driver which I run on Win 2008 Server and on Win 2012 Server. When the Que-depth is increased from 64 to 256 there is no reduction in performance on Win 2008 but reduces considerably on Win 2012. Vtune analysis on both systems give prettymuch the same results.
When the Que-depth is 64 the driver performs about 720K I/Os per second
and when Que depth is increased to 256 there is a reduction to 620K I/Os per second
Vtune indicated some functions which is taking time,even after ixing them it is still performing bad.
It is related to USB performance measurement, but it give some info about the commands and switches of Xperf and Logman.
BTW, have you tried General Exploration analysis in VTune? It may highlight some HW issues (look to red cells).
iliyapolak, can you please share your thoughts which useful XPerf features are missing in VTune?
I think that Xperf models or measures betters Win OS environment and its interaction with the software,although it is not probably the main purpose of the VTune.
For example time CPU spent in DPC routine.