Profiling Generated Code

[lightningth]

I'm working on a project that generates x86 code that is then executed in memory. VTune picks up on this processing and does indicate how much time is spent in this area. I am wanting to get further information on what areas are slowing down, however, the code is generated on the fly so I do not have an executable to point VTune to for disassembly to get timing data. I can dump the memory chunk to a file, but if I do that, what is required to get VTune to properly load it so I can see timing data?

[Huseyin_A_Intel1]

Hi,

i have just answered a question which might be a help to you?

http://software.intel.com/en-us/forums/showthread.php?t=60762

" In case you have a way of modifying the JIT compiler (or the code generated) you can leverage the JIT Profiling API that we will releasewith VTune Analyzer 9.1 for Windows and Linux. The VTune Performance Analyzers JIT (Just-In-Time) Profiling API will provide functionality to profile runtime generated code and fix the problems you are encountering. This API allows analysis of runtime generated code with both sampling and call graph profilers which are already available with VTune Performance Analyzer. The VTune Analyzer 9.1 will become available soon. If you think this might help you, please let me know and I can share more information."

I hope this answer also helps you.

Thanks

Levent