Kilian -

APEX+OTF should work fine with distributed runs.  Are you running the job from 
a directory where you have write permission?  Here are some tips -

- OTF requires a "unification" process at the end of the run.  Ideally, I would 
use HPX to do that, but I haven't implemented that.  Instead, I use the 
filesystem.  So, each rank will write three (or so) "hidden" files in the 
current working directory, and rank/locality 0 will read them, unify them, and 
write out a "master" mapping for the other ranks.

- By default, APEX will write the OTF2_archive directory of OTF2 files in the 
current working directory.

- If you want to change the location of either of those two locations, set the 
APEX_OTF2_ARCHIVE_PATH environment variable before your run, and the files will 
be written there.  The default value for APEX_OTF2_ARCHIVE_PATH is "." (current 
working directory, relative).

Let me know if that works, and/or if you have any other problems.

Thanks -
Kevin

> On Mar 8, 2019, at 4:44 AM, Kilian Werner <[email protected]> wrote:
> 
> Thank you very much Kevin!
> 
> It works flawlessly and is a very handy profiling tool.
> 
> It is inteded to work within distributed settings to right?
> As soon as I run the application with mpirun e.g. with multiple localities 
> apex complains:
> 
> [OTF2] src/otf2_archive_int.c:1108: error: Unknown error code: Couldn't 
> create directories on root.
> OTF2 Error: INVALID, Unknown error code
> 
> 
> While I do not have root access while running the application, I do have 
> writing rights for the working directory (thus the directory where the 
> OTF2_archive is put).
> 
> Am I missing some additional steps for running apex otf2 support on multiple 
> locations, is it not inteded to support that or is the problem probably 
> located within my toolchain and environment?
> 
> Thanks,
> 
> Kilian
> 
> On Tue, 26 Feb 2019 10:13:12 -0800
> Kevin Huck <[email protected]> wrote:
>> Kilian -
>> You don't need TAU to generate OTF2 traces directly from APEX.  Here is what 
>> you need to generate them:
>> Get and build the OTF2 library (straightforward configure/make process):
>> https://www.vi-hps.org/projects/score-p/
>> Configure HPX with APEX enabled and OTF2 support enabled:
>> cmake <usual hpx options...> -DHPX_WITH_APEX=TRUE -DAPEX_WITH_OTF2=TRUE 
>> -DOTF2_ROOT=/path/to/otf2 ...
>> If you want to make sure you have the most recent APEX, you can also add:
>> -DHPX_WITH_APEX_NO_UPDATE=TRUE -DHPX_WITH_APEX_TAG=develop
>> ...but I think the current HPX master uses APEX 2.1.2, which is up to date.
>> After building HPX and your application, set the following environment 
>> variable before executing:
>> APEX_OTF2=1
>> You should see an OTF_archive directory after the run, you can load the 
>> APEX.otf2 file in vampir.
>> Let me know if you run into any problems -
>> Kevin
>>> On Feb 26, 2019, at 5:16 AM, Hartmut Kaiser <[email protected]> 
>>> wrote:
>>> Kilian,
>>> I am cc'ing Kevin Huck to make sure he sees this. He is Mr. APEX and will be
>>> able to give you all the information you need.
>>> Regards Hartmut
>>> ---------------
>>> http://stellar.cct.lsu.edu
>>> https://github.com/STEllAR-GROUP/hpx
>>>> -----Original Message-----
>>>> From: [email protected] <hpx-users-
>>>> [email protected]> On Behalf Of Kilian Werner
>>>> Sent: Tuesday, February 26, 2019 3:03 AM
>>>> To: [email protected]
>>>> Subject: [hpx-users] HPX Profiling (TAU?)
>>>> Dear list!
>>>> I am trying to optimize my HPX project and was looking for a way to
>>>> observe per-thread traces of the execution (in a vampir trace like
>>>> fashion).
>>>> I am already successfully utilizing APEX Support, however to my
>>>> understanding this only yields total times/percentages.
>>>> I did find references to HPX supporting TAU through APEX, however the
>>>> HPX_WITH_TAU flag was removed in
>>>> https://github.com/STEllAR-GROUP/hpx/pull/2380
>>>> Was TAU support dropped or did the process of building hpx with tau
>>>> change? If so, how would I go about building it.
>>>> If it was dropped are there any other ways to trace HPX in a vampir trace
>>>> / gantt-chart like fashion?
>>>> Best regards,
>>>> Kilian Werner
>>>> _______________________________________________
>>>> hpx-users mailing list
>>>> [email protected]
>>>> https://mail.cct.lsu.edu/mailman/listinfo/hpx-users
>> --
>> Kevin Huck, PhD
>> Research Associate / Computer Scientist
>> OACISS - Oregon Advanced Computing Institute for Science and Society
>> University of Oregon
>> [email protected]
>> http://tau.uoregon.edu
> 

--
Kevin Huck, PhD
Research Associate / Computer Scientist
OACISS - Oregon Advanced Computing Institute for Science and Society
University of Oregon
[email protected]
http://tau.uoregon.edu







_______________________________________________
hpx-users mailing list
[email protected]
https://mail.cct.lsu.edu/mailman/listinfo/hpx-users

Reply via email to