Re: endrun(888) problem

From: Volker Springel <volker_at_MPA-Garching.MPG.DE>
Date: Wed, 7 May 2014 21:56:15 +0200

I think Ali's hint is spot on. The simulation is likely so small that the ratio between the number of tree nodes required by the code and the particle number is much larger than normally needed for well-loaded processors. It should be possible to cure this with a (drastic) increase of the parameter 'TreeAllocFactor' in the parameter file.

Volker

On May 7, 2014, at 5:46 PM, Ali Snedden wrote:

> You might read the user manual and check that you are using reasonable values in your parameter file (specifically All.PartAllocFactor). Perhaps All.MaxPart is too small or you're not using a large enough value for MaxNodes. Good luck.
>
>
> ~ali
>
>
> On Wed, May 7, 2014 at 11:31 AM, arlindo.trindade_at_gmail.com <arlindo.trindade_at_gmail.com> wrote:
> Hi Ali,
>
> Thanks for your email.
>
> I've changed the values that are passed to each endrun as you suggested and the code breaks at line 371 in the forcetree.c file. But still I can't find out what causes this error.
>
> Do you have any ideas?
>
> Best regards,
> Arlindo
>
>
>
> 2014-05-07 16:03 GMT+01:00 Ali Snedden <asnedden_at_nd.edu>:
>
> Hello Arlindo,
>
> It would be nice to know which line it actually breaks at. You could use a debugger like Gdb or just change the values passed to each line that has endrun(888).
>
> It is worth your time to learn how to use a debugger. >From my personal experience, using a debugger has probably cut my development time between 30-50% compared with just using print statements.
>
> To use Gdb for a parallel program. First compile Gadget with the '-g' compiler option. Then enter the following into your command line.
>
> mpirun -np 2 xterm -geometry 100x62 -sb -sl 10000 -e gdb ./Gadget2
>
> I added a bunch of extra options to customize the x11 window. Then in each of your two x11 windows enter the command line arguments (i.e. pass your parameter file).
>
> start lcdm.param
>
> Then you can set breakpoints and do other neat things by following these helpful websites.
>
> http://betterexplained.com/articles/debugging-with-gdb/
> http://www.unknownroad.com/rtfm/gdbtut/gdbtoc.html
>
>
> Best Regards
> ~Ali
>
>
> On Wed, May 7, 2014 at 10:42 AM, Arlindo Trindade _at_ gmail <arlindo.trindade_at_gmail.com> wrote:
>
> Hi all,
>
> I'm trying to running some N-body simulation tests with Gadget 2 on a cluster. The simulation is very very small, the number of particles is N=2*16^3 ( dark matter + gas) and the boxsize is L=3.125 Mpc. However I get a endrun 888 error. The same thing happens if I increase the size of the simulation ( both L and N). I've identified the files where the function endrun is called with the code 888 ( timestep.c lines 482 and 530, forcetree.c line 371) but I still can't understand what the problem is and thus I can't solve this problem.
>
> Does anyone has a suggestion?
>
> Cheers,
> Arlindo
>
>
>
>
> -----------------------------------------------------------
> If you wish to unsubscribe from this mailing, send mail to
> minimalist_at_MPA-Garching.MPG.de with a subject of: unsubscribe gadget-list
> A web-archive of this mailing list is available here:
> http://www.mpa-garching.mpg.de/gadget/gadget-list
>
>
>
> -----------------------------------------------------------
>
> If you wish to unsubscribe from this mailing, send mail to
> minimalist_at_MPA-Garching.MPG.de with a subject of: unsubscribe gadget-list
> A web-archive of this mailing list is available here:
> http://www.mpa-garching.mpg.de/gadget/gadget-list
>
>
>
>
> --
> Arlindo Trindade
>
>
> -----------------------------------------------------------
>
> If you wish to unsubscribe from this mailing, send mail to
> minimalist_at_MPA-Garching.MPG.de with a subject of: unsubscribe gadget-list
> A web-archive of this mailing list is available here:
> http://www.mpa-garching.mpg.de/gadget/gadget-list
>
>
>
> -----------------------------------------------------------
>
> If you wish to unsubscribe from this mailing, send mail to
> minimalist_at_MPA-Garching.MPG.de with a subject of: unsubscribe gadget-list
> A web-archive of this mailing list is available here:
> http://www.mpa-garching.mpg.de/gadget/gadget-list
Received on 2014-05-07 21:56:17

This archive was generated by hypermail 2.3.0 : 2022-09-01 14:03:42 CEST