Have you tried to use the performance tool "perfsys"provided in Intel IPP? How have your found it useful?
The Intel IPP (both for Windows* and Linux*) on Intel Pentiumand Itanium processorsprovides a great powerful tool "perfsys" toolto test theperformance of each function in the API.
Once you fully install the Intel IPP package, you can find the perfsys tool located in the ipp40 oolsperfsys directory. For instance, you can simply enter ps_ipps in the DOS command prompt to check the performance data for signal processing primitives (ippSP), or type ps_ipps -h to list the options for running this .exe file.
If you want to measure certain functions (for example: FIRLMS_32f) instead of a whole set of ippSP functions, you can enter the following command at a DOS Prompt:
ps_ipps -fFIRLMS_32f -V firlms.csv
(run with "-h" to get a short list of possible parameters)
After running the test, a file named firlms.csv will be created. Next, use Microsoft Excel to view this file that contains both the elapsed execution time in microseconds, as well as in the more commonly used units, clocks per element (cpe).
Please go tothisweb site: http://support.intel.com/support/performancetools/libraries/ipp/ia/perfsys.htmto get more information on howto interpret the performance data.
Hope it helps.