flex-em in parallel mode?
Hi,
I am trying to run flex-em.py out of the cryoem_files examples in parallel mode.
I tried for example the following script. It runs. Nevertheless only one CPU is used. (The example model-parallel.py works perfectly)
---------------------------------------------------------------------------------------------------- from modeller import * from MD import opt_md from modeller.parallel import *
log.minimal() j = job()
j.append(local_slave()) #CPU 1 j.append(local_slave()) #CPU 2
log.verbose() # request verbose output env = environ()
j.queue_task(opt_md('./', 'xxxx', str(2), 10*2, 'mod2b.mrc', 'mod2bn.pdb', 'MRC', 1.8, 64, 15, 1.8, 1.8, 1.8, 'rigid.txt'))
results= j.run_all_tasks()
---------------------------------------------------------------------------------------------------------
Thanks for your help Nathalie
On 10/19/2010 09:01 AM, Nathalie Braun, TUM wrote: > I am trying to run flex-em.py out of the cryoem_files examples in > parallel mode.
The FlexEM protocol has not been designed to run in parallel. However, if you have multiple systems to model you could run multiple copies of FlexEM simultaneously (in different directories) to get the same effect.
> I tried for example the following script. It runs. Nevertheless only one > CPU is used.
The parallel framework distributes tasks between available processors. Each task uses a single processor - tasks are not split across processors. In your script you only queue a single task, so only one of your two CPUs is used. If you queue two or more tasks, both CPUs will be utilized.
Ben Webb, Modeller Caretaker
participants (2)
-
Modeller Caretaker
-
Nathalie Braun, TUM