[hpct] Summary of the issues: Sept 4, 2008
Vitali A. Morozov
morozov at anl.gov
Wed Sep 3 16:23:41 CDT 2008
Hi Guojing;
Thank you very much for a quick response. I am also copying this message
to the list so that everybody know we have been working on some
mentioned issues.
1) I confirm, 5 is fixed. Great job!
2) I confirm, 6 is fixed. Very good!
A related question is how to search for a second occurrence?
3) We had the histogram capability in previous releases, so why cannot
we continue having it? What has changed since, say, March?
4) Please take a look at the picture you sent me in the next message:
there are two scroll bars: internal, which is what you are looking at,
and external, which never scrolls.
Issue 1) Agree - difficult to repeat. But there is second part - broken
PostScript output.
Issue 2) Open
Issue 3) In discussion
Issue 4) Open
Issue 5) Fixed
Issue 6) Fixed
Basically we have 3 issues, and one discussion.
Thanks again,
Vitali
Guojing Cong wrote:
>
> hi Vitali,
>
>
> I believe I fixed 5, 6, and 3. Note that the histogram capability
> would have to depend on another software that we (IBM) do not own.
> Besides, the original Xprofiler on AIX does not have this capability
> either. If you insist on having that, we can open another chanel for
> discussion, and see how we might get that for you. There is no
> technical problem, the issue I believe is purely legal.
>
> Now for 1, as it is sporadic, we will need to find a way to reproduce
> that in a repeatable manner. It might take more time to fix. For 4,
> there is a partial fix, and we are working to get it in there. For 2,
> I am investigating. I remember the scroll bar is used when there are
> many function calls. I will send you a separte note with the
> xprofiler attached.
>
> Regards
> Guojing
>
>
>
>
>
>
>
>
>
>
> *"Vitali A. Morozov" <morozov at anl.gov>*
> Sent by: hpct-bounces at lists.alcf.anl.gov
>
> 09/03/2008 04:18 PM
>
>
> To
> HPCT Mailing list <hpct at alcf.anl.gov>
> cc
>
> Subject
> [hpct] Summary of the issues: Sept 4, 2008
>
>
>
>
>
>
>
>
>
> Sept 4, 2008: HPCT: Summary of the issues
>
> 1) We want to be able to reduce statistics, generated in mpi_profile.*
> files, by some sort of MPI_Start_Statistics()/MPI_Stop_Statistics()
> function calls. As we know from I-hsin,
> MPI_trace_start()/MPI_trace_stop() calls control tracing part only.
>
> 2) According to I-hsin, "mpi_profile.<rank number> are generated per
> each MPI rank." We do not see this behavior: 4 files are generated
> instead with minimal, maximal, mean communication time, and collective
> over all tasks statistics file called mpi_profile.0. Please clarify 1)
> what is printed at the end of the run, 2) how to print statistics from
> a particular MPI task. Examples of the test run would be very useful
> to have.
>
> 3) What is IHPCT_BASE env. variable? When I should use it?
>
> 4) peekperf does not show sources for Fortran program.
>
> 5) Fortran instrumented program generates "Error getting the
> executable name, line number information will not be available" message.
>
> 6) peekview single_trace gives the node 0, 8, and 15. Example program
> was sent to the list already and contains 2 Sends and 2 Recvs. Where
> are the other communications? See attached screenshot.
>
> 7) peekview on selective instrumentation does not show correct number
> of calls. From 1 Send and 1 Recv, we see nothing on peekview screen.
>
> 8) HPM part does not work: core file at runtime. Please, give us
> example of 1) how to compile HPM-instrumented program, 2) what is the
> result of the run.
>
> 9) Xprofiler has 6 open issues:
> 1. Load Files, Filter -> Uncluster functions, File -> Screen
> Dump -> Select Target Window,
> 1. clicking now on the main window may occasionally cause crach
> 2. resulting file Xprofiler_screenDump.ps.0 is a balack-box
> Postscript
> 2. The meaning on exterior scroll bars in not clear, they never
> scroll.
> 3. *Removing Histogram functionality is unacceptable* A more
> complicated multistep example
> 1. Load Files
> 2. Report -> Flat Profile
> 3. Code Display -> Show Source Code
> 4. File -> Save as; File -> Close
> 5. Code Display -> Show Histogram w/Source Code
> 6. crash
> 4. Load Files, Report -> Flat Profile, Code Display -> Show
> Source Code, File -> Save as
> 1. Close the window by clicking X (not File -> Close)
> 2. Select another function in Flat Profile
> 3. Code Display -> Show Source Code; File -> Save as; crush
> 5. Load Files, Report -> Flat profile, Show Source Code, search
> anything non-existent, press enter, crash
> 6. Load Files, Report -> Flat Profile, try to search anything
> like 'asfsafd', once not found press Enter, crash
>
> Vitali
> _______________________________________________
> hpct mailing list
> hpct at lists.alcf.anl.gov
> http://lists.alcf.anl.gov/cgi-bin/mailman/listinfo/hpct
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alcf.anl.gov/pipermail/hpct/attachments/20080903/e010c6c5/attachment.htm>
More information about the hpct
mailing list