Evidently the place to change the number of processes is here:
%%bash# start simulation./simfactory/bin/sim create-submit tov_ET \  
--parfile=par/tov_ET.par --procs=14 --num-threads=14 --walltime=0:20:0
At first I didn't see anything happening, but the CPU was getting hot (about 
65C), so I suspected that something was going on, and was able to obtain the 
star plot. Perhaps I can now follow some of the erudite discussions in this 
forum.
I also prefer to use Linux natively for such projects, as I did here (Ubuntu).





 
 

    On Monday, October 19, 2020, 09:50:52 AM PDT, Roland Haas 
<[email protected]> wrote:  
 
 Hello Robert,

does it actually abort? Symmetric multiprocessing is tricky to get
right automatically and the automated CPU detection in simfactory that
runs during setup-silent is designed to basically ignore them.

The HelloWorld code is very simple and does not make any attempt at
getting it "right". As long as you get only "Warnings" there is nothing
to worry about.

"Real" simulations usually use the "hwloc" and "SystemTopology" thorns
which query the CPU and do a (much) better job at laying out threads
onto physical cores and hardware threads. Codes in Cactus are typically
of types where SMT gives only minimal benefit so is usually only
worthwhile on big clusters and large production runs but not on
workstations used for testing and debugging so the automated CPU
detection code is simfactory does not handle it (or at least chooses to
err on the side of caution).

Yours,
Roland

> I have run CactusTutorial.ipynb and obtained output for Hello World. However, 
> when the characteristics of my computer (i9-7940X) were guessed, apparently 
> something went wrong. I obtained the following output:
> Warning: Too many threads per process specified: specified num-threads=14 
> (ppn-used is 14)Warning: Total number of threads and number of threads per 
> process are inconsistent: procs=1, num-threads=14 (procs*num-smt must be an 
> integer multiple of num-threads)Warning: Total number of threads and number 
> of cores per node are inconsistent: procs=1, ppn-used=14 (procs must be an 
> integer multiple of ppn-used)+ set -e+ cd 
> /home/robert/simulations/helloworld/output-0000-active+ echo Checking:+ pwd+ 
> hostname+ date+ echo Environment:+ export CACTUS_NUM_PROCS=1+ export 
> CACTUS_NUM_THREADS=14+ export GMON_OUT_PREFIX=gmon.out+ export 
> OMP_NUM_THREADS=14
> I can reduce the number of threads, but I'd rather not. I don't know enough 
> yet about how this software works to set procs, though: it is not explicitly 
> mentioned as such in the machine database. How would I do this?
> 
> 
> 
>  


-- 
My email is as private as my paper mail. I therefore support encrypting
and signing email messages. Get my PGP key from http://pgp.mit.edu .  
_______________________________________________
Users mailing list
[email protected]
http://lists.einsteintoolkit.org/mailman/listinfo/users

Reply via email to