Paul Taylor wrote:
> Hi-
>
> I've got a simulation where I change particle types to simulate
> accretion; P[i].Type=0 particles go to P[i].Type=5 particles when they're
> accreted, and then I reorder the lists so that all the N_gas
> (SPH) particles are first, followed by the Type5 ones. That all seems to
> be working ok, I'm calling the conversion/rearranging in the run.c file,
> between the every_timestep_stuff() and domain_Decomposition() calls.
> However, the process will just get stuck after a particle has been
> converted/rearranged, and using the the gdb to see what's happening, it's
> getting stuck at this point:
>
> Program received signal SIGINT, Interrupt.
> 0x5590d3a1 in gettimeofday () from /lib/tls/libc.so.6
>
> (the first line is me interrupting the process). Other people I know have
> had this problem as well. I have no idea why it would have to do with
> gettimeofday... Any suggestions?
It is unclear to me how the hang in gettimeofday() could arise. Maybe
you are fooled somehow by gdb when you attach to the running process (I
assume that's what you're doing). Sometimes this happens as a result of
aggressive compiler optimizations. I'd therefore suggest to switch them
off completely, and then try again.
My suspicion is that the code gets stuck in the tree walk somewhere as a
result of your particle rearrangement. If you do this, the tree becomes
invalid and you need to force the code to construct a new one. You can
do this by adding a statement of the form
All.NumForcesSinceLastDomainDecomp = 1 +
All.TreeDomainUpdateFrequency * All.TotNumPart;
after you made the rearrangement, and before calling the domain
decomposition.
Volker
>
> Cheers,
> PT
>
>
>
>
> -----------------------------------------------------------
>
> 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 2007-03-13 18:53:15