Erik
Actually, even with --procs=2 --num-threads=1 I will get: The environment variable CACTUS_NUM_PROCS is set to 2, but there are 1 MPI processes.
However, with only --procs=2 can really start a simulation, and terminate due to TAT/Slab limitation
Is there any configuration corresponding to "but there are 1 MPI processes. This may indicate a severe problem with the MPI startup mechanism."
I'm using optionlist ubuntu.cfg submitscript generic.sub runscript debian.run
Cory
On Fri, Dec 23, 2016 at 9:44 PM, Erik Schnetter schnetter@cct.lsu.edu wrote:
Cory
Using "--procs=4 --num-threads=1" might help. You do not need to change the settings in the ".ini" file; only the command line parameters determine how many MPI processes are used.
-erik
On Fri, Dec 23, 2016 at 8:12 AM, Cory Chu cory.yk.chu@gmail.com wrote:
Hi Ian
I'am running on VirtualBox set whit 2 Processors and 8GB RAM
When i try with this submit command
create-submit mybbh --parfile par/BBHLowRes.par --num-threads=1 --procs=2
I get following message in .err file
WARNING level 0 from host kpc-VirtualBox process 0 while executing schedule bin (none), routine (no thorn)::(no routine) in thorn Carpet, file /home/kuangtly/programming/ET/forfun/Cactus/arrangements/
Carpet/Carpet/src/SetupGH.cc:153:
-> The environment variable CACTUS_NUM_PROCS is set to 2, but there
are 1
MPI processes. This may indicate a severe problem with the MPI startup mechanism. cactus_sim: /home/kuangtly/programming/ET/forfun/Cactus/arrangements/
Carpet/Carpet/src/helpers.cc:275:
int Carpet::Abort(const cGH*, int): Assertion `0' failed. Rank 0 with PID 15985 received signal 6 Writing backtrace to BBHLowRes/backtrace.0.txt Aborted (core dumped)
i have change these setting in .ini file ppn =2 max-num-threads =2 num-threads =2
On Fri, Dec 23, 2016 at 6:28 PM, Ian Hinder ian.hinder@aei.mpg.de
wrote:
On 23 Dec 2016, at 10:52, Cory Chu cory.yk.chu@gmail.com wrote:
Erik
Should I simply change the .ini file in simfactory/mdb/machines with ppn = 2 max-num-threads = 2 num-threads = 2
and adding --procs = 2 on running command ?
sorry i just can not find a detail documentation on this part after i change the above setting the problem did not solved
Hi Cory,
Can you give the command line you are using to submit the simulation?
You
need to ensure that the number of MPI processes you use is larger. You could do this by running fewer threads per process, or if you are
already
running one thread per process, by increasing the number of processes by increasing the number of threads. That might mean reconfiguring the
machine
definition to allow more cores to be used than you actually have. What machine is it; i.e. how many processors and cores do you have?
-- Ian Hinder http://members.aei.mpg.de/ianhin
-- Erik Schnetter schnetter@cct.lsu.edu http://www.perimeterinstitute.ca/personal/eschnetter/