On 3 May 2017, at 05:23, Robert Kopp <[email protected]> wrote:
> The message I sent a little earlier with a file attachment may not reach the > list because of its size, 600K or more (includes all the data points in the > simulation). Perhaps this will be enough to determine the source of the > problem, though--at least, for someone with experience with this application. > Otherwise I'll try to reduce the size of the attachment to <60K and find the > relevant portions. > > INFO (Carpet): MPI is enabled > INFO (Carpet): Carpet is running on 1 processes > INFO (Carpet): This is process 0 > INFO (Carpet): OpenMP is enabled > INFO (Carpet): This process contains 1 threads, this is thread 0 > INFO (Carpet): There are 1 threads in total > INFO (Carpet): There are 1 threads per process > INFO (Carpet): This process runs on host robert-desktop, pid=17940 > INFO (Carpet): This process runs on 8 cores: 0-7 > INFO (Carpet): Thread 0 runs on 8 cores: 0-7 > INFO (Carpet): This simulation is running in 3 dimensions > _______________________________________________ > Users mailing list > [email protected] > http://cactuscode.org/mailman/listinfo/users Hi Robert, Can you post your machine definition file? This is the file that was created when you ran "sim setup". As Roland said, simfactory sets OMP_NUM_THREADS itself, so whatever you set in the shell in which you invoke simfactory will be ignored. Assuming you want a single process with 4 threads, you should use --procs 4 --num-threads 4 to the submission command line. This means "run on 4 cores in total, and use 4 threads per process". Yes, the terminology is strange. The defaults for these are stored in your machine definition file. I don't think there is any attempt made to default to using all the threads, but I might be wrong. -- Ian Hinder http://members.aei.mpg.de/ianhin _______________________________________________ Users mailing list [email protected] http://cactuscode.org/mailman/listinfo/users
