From: Axel Kohlmeyer (akohlmey_at_gmail.com)
Date: Tue Jul 16 2013 - 14:11:35 CDT
On Tue, Jul 16, 2013 at 9:05 PM, Adrian <adpala_at_hotmail.com> wrote:
> Hello,
>
> We have a workstation with 12 CPU's and 4 GPU's and we are trying to run
> NAMD simulations in it, but we are not sure we are actually using the GPU's.
>
> Here is the first part of a log file using the workstation.
>
> Charm++> scheduler running in netpoll mode.
> Charm++> Running on 1 unique compute nodes (12-way SMP).
> Charm++> cpu topology info is gathered in 0.007 seconds.
> Info: NAMD 2.8 for Linux-x86_64-CUDA
> Info:
> Info: Please visit http://www.ks.uiuc.edu/Research/namd/
> Info: for updates, documentation, and support information.
> Info:
> Info: Please cite Phillips et al., J. Comp. Chem. 26:1781-1802 (2005)
> Info: in all publications reporting results obtained with NAMD.
> Info:
> Info: Based on Charm++/Converse 60303 for net-linux-x86_64-iccstatic
> Info: Built Sat May 28 11:30:15 CDT 2011 by jim on larissa.ks.uiuc.edu
> Info: 1 NAMD 2.8 Linux-x86_64-CUDA 4 cabeza.compbiophyslab.com Guest
> Info: Running on 4 processors, 4 nodes, 1 physical nodes.
> Info: CPU topology information available.
> Info: Charm++/Converse parallel runtime startup completed at 0.00935698 s
> Pe 3 physical rank 3 binding to CUDA device 3 on cabeza.compbiophyslab.com:
> 'Tesla C2070' Mem: 4095MB Rev: 2.0
> Pe 1 physical rank 1 binding to CUDA device 1 on cabeza.compbiophyslab.com:
> 'Tesla C2070' Mem: 4095MB Rev: 2.0
> Did not find +devices i,j,k,... argument, using all
> Pe 0 physical rank 0 binding to CUDA device 0 on cabeza.compbiophyslab.com:
> 'Tesla C2070' Mem: 4095MB Rev: 2.0
> Pe 2 physical rank 2 binding to CUDA device 2 on cabeza.compbiophyslab.com:
> 'Tesla C2070' Mem: 4095MB Rev: 2.0
these lines tell you, that you are.
> Info: 1.63564 MB of memory in use based on CmiMemoryUsage
> Info: Configuration file is
> /home/Guest/Adrian/asyn/ProductionNa15/runNa15.namd
> Info: Changed directory to /home/Guest/Adrian/asyn/ProductionNa15
> ......................................................................................
>
> Which is the correct command to run the NAMD in the workstation?
>
> We are currently using one similar to this:
>
> charmrun ++local +p4 namd2 +idlepoll [configuration file] > [log file]
you can also try using more than 1 CPU per GPU. try +p8 and +p12
running nvidia-smi should tell you how well the GPUs are utilized and
comparing running a non-GPU executable vs. the executable with CUDA
support should tell you how much speedup you have. that will strongly
depend on many factors
(hardware and simulated system), so it is not possible to make any
general comments on that.
you may want to search through the mailing list archives for several
discussions on the subject. some of them quite long-winded and
detailed. no need to repeat all of that... ;-)
cheers,
axel.
>
>
>
> Thanks for your time!
> Adrian Palacios
>
-- Dr. Axel Kohlmeyer akohlmey_at_gmail.com http://goo.gl/1wk0 International Centre for Theoretical Physics, Trieste. Italy.
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:26 CST