[phenixbb] nproc for autobuild

Thomas C. Terwilliger terwilliger at lanl.gov
Wed Nov 9 06:32:05 PST 2011


Hi Mike,
Yes, you have all of that correct.  If you are building 10 models in
parallel, you can use 50 processors effectively.  With
rebuild-in-place=true, normally you can use 5 processors for each model
you are building in parallel.
All the best,
Tom T

>> Hello,
>>
>> I apologize if this question seems very basic, but I am new to
>> phenix.autobuild. I would like to set up a series of phenix.autobuild runs
>> on our cluster, and would like to maximize the speed.  I am trying to make
>> single rebuilt models using the command below:
>>
>>
>> phenix.autobuild data=data.mtz model=coords.pdb map_file=map_coeffs.mtz
>> multiple_models_number=1 n_cycle_rebuild_max=5 ncycle_refine=5
>> s_annealing=true refine_with_ncs=True place_waters=True nproc=??
>> run_command=/home/qsub.sh
>>
>>
>> The AutoBuild documentation says "if you are using rebuild_in_place=False,
>> then use nproc=4" and "if you are using rebuild_in_place=True, then use
>> nproc=5. (Any more will not make any difference)." My interpretation of
>> the documentation is that for the above AutoBuild run with
>> multiple_models_number=1, I should set nproc=5 (or 4?). Is this correct?
>> Also, if I were to change the command so that multiple_models_number=10,
>> should I then use nproc=50 (i.e. 5 processors per model), or is it still
>> the case that setting nproc>5 doesn't make a difference?
>>
>> Thanks in advance for your help!
>>
>> Mike
>>
>>
>>
>> --
>> Michael C. Thompson
>>
>> Graduate Student
>>
>> Biochemistry & Molecular Biology Division
>>
>> Department of Chemistry & Biochemistry
>>
>> University of California, Los Angeles
>>
>> miket at chem.ucla.edu
>> _______________________________________________
>> phenixbb mailing list
>> phenixbb at phenix-online.org
>> http://phenix-online.org/mailman/listinfo/phenixbb
>>



More information about the phenixbb mailing list