Here all the files are in one .pdf, which should be easier to read. (It's about
40K, which should be acceptable, but .odt is even smaller, if it's necessary to
use that instead.)
On Sunday, May 7, 2017 11:55 PM, Ian Hinder <[email protected]> wrote:
On 8 May 2017, at 04:43, Robert Kopp <[email protected]> wrote:
I had to change my posting email address because I was getting too many bounces
at the previous one. So apparently I have to start a new thread.
Changing "procs" to 4 is apparently understood to mean changing nodes to 4. (I
think a node is a device that has or can have an IP--anyhow, it does--but I
intend it to mean instead that four threads are supported.) Perhaps this will
indicate what the problem is, though.
robert@robert-desktop:~/Cactus$ ./simfactory/bin/sim submit static_tov
--parfile=par/static_tov_small.par --procs=4 --walltime=8:0:0Warning: job
status is UWarning: job status is UAssigned restart id: 3 Error: Too many nodes
specified: nodes=4 (maxnodes is 1)Aborting
Simfactory._______________________________________________
Users mailing list
[email protected]
http://cactuscode.org/mailman/listinfo/users
Hi Robert,
It's not interpreting --procs as the number of nodes to use. It's taking all
the configuration options and the options you pass into account, and trying to
work out what you mean (there are a large number of possibilities for how to
handle the parallelisation, and it's a bit complicated to make an interface
which is powerful enough to allow all possibilities, and also make it simple in
simple cases).
However, I don't understand what is happening now. Now that you have changed
to using --procs 4, please can you send the files again, since they will have
changed? SimFactory should be calculating the number of nodes based on the
number of cores you ask for, and the number of cores per node which, if you
have "ppn = 4" in your ini file, and --procs 4 on the command line, should give
you one node. The properties.ini files from the simulation will reveal what
simfactory has computed for each setting.
-- Ian Hinderhttp://members.aei.mpg.de/ianhin
_______________________________________________
Users mailing list
[email protected]
http://cactuscode.org/mailman/listinfo/users
