- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello,
I'm trying to analyze MPI applications with Intel SDE and generate dynamic control-flow graph data for the compute regions only with the following command:
sde64 -disasm_att 1 -dcfg 1 -dcfg:write_bb 1 -emu_fast 1 -start_ssc_mark 111:repeat -stop_ssc_mark 222:repeat ... <few more ops> -- <app>
Without start_ssc_mark and stop_ssc_mark, I'm getting one .bb.txt and one .json file per MPI rank, but it includes undesired MPI calls, so I created a PMPI wrapper to inject __SSC_MARK before and after MPI calls to filter them.
However, now I'm getting for each region (likely related to those https://software.intel.com/content/www/us/en/develop/articles/pintool-regions.html) a separate bb.txt and json file, and hence the number of output files is 2*(#mpi_calls+1)*#ranks.
Is there a way to tell SDE to merge those somehow back into 1 file each and not dump terabytes and millions of files?
Best,
JD
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
Thanks for reaching out to us.
Since the issue related to Intel SDE, we are moving this query to Intel ISA Extensions forum for a faster response.
Thanks.
![](/skins/images/7B0AB6865064EAE32B30EC9A4E94B48A/responsive_peak/images/icon_anonymous_message.png)
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page