I know that there were some questions about attaching the Intel XE Inspector to an active process (PID) at certain point from the start time.
I wanted to ask wheather there is any update from the Intel guys on that regard in one of the updates of the Inspector XE ?
If not, what are your kind suggestion for debugging certain source code file for data race within a large application ? Do I have to wait until the Inspector will reach the parallel area ?
Thanks in advance,
There is not any way to attach to a running process, however, if the problem is just that you are only interested in (and therefore only want Inspector to be instrumenting) a small portion of the code, there are a couple of ways that you can do that.
One is to use the collection control API to turn on and off analysis. You can find out more about that at https://software.intel.com/sites/products/documentation/doclib/iss/2013/inspector/lin/ug_docs/GUID-0F449BFC-2B91-4B0D-A990-211A8538145D.htm or in the API section of your installed help. (Functionality added in 2013 version).
You could also take advantage of the debugger integration. Use the "delay analysis until breakpoint" as described in https://software.intel.com/en-us/articles/intel-inspector-xe-2013-integrated-debugger-support .
Hopefully one of these will work for you.
@ Holly Wilper - Thanks for the information.
I kindly have a follow-up question : If I want to analyze a certain routine in the WRF model which consist OpenMP directives - do you know what flags and where I defined them before compilation (obviously, configure.wrf but where exactly) ?
Is there any way to increase the memory limits of the Inspector XE ? The other day, I have analyzed large application (a model) and an error of memory usage limit was shown at the main interface (probably because it uses numerous COMMONS for sharing data) ; I want to know how to prevent this when analyzing large models like WRF.
Thanks very much,