Re: Two GPU-based workstation

From: James Starlight (jmsstarlight_at_gmail.com)
Date: Sun Nov 03 2013 - 02:38:59 CST

updating

using namd2 +idlepoll +p4 +devices 0,1 ./restart.conf
I've launched simulation on both GPUs (according to thermal monitoring in
nvidia-settings) but only half of cpus were fully loaded.

By the way how I could monitor real GPU loading as well as namd performance
( in ns\days or GFlops )?

James

2013/11/1 James Starlight <jmsstarlight_at_gmail.com>

> Ok. I'll try to make some simulations of this configure. The main issue
> with which I can force is the possible conflict between that older cuda
> library (used from vmd) and more newest development driver ( 5.5 version)
> which comes from installed cuda-5.5.
>
> By the way how I could use both of the GPUs simultaneously ? Just use the
> below command?
>
> namd2 +idlepoll +p4 +devices 0,1 ./restart.conf
>
> Where 0 and 1 are the ids of my GPUs? Is there additional options for
> synchronization of the simulations in dual-GPU regime ?
>
> James
>
>
> 2013/10/31 Aron Broom <broomsday_at_gmail.com>
>
>> don't replace anything, just point to the version of the library in your
>> NAMD directory as you did. It should work fine.
>>
>>
>> On Thu, Oct 31, 2013 at 1:24 PM, James Starlight <jmsstarlight_at_gmail.com>wrote:
>>
>>> Dear Namd users,
>>>
>>> I've build my new workstations consisted of two Titans with i6 (linux
>>> recognize it like 12 core process but actually it consist of 6 nodes).
>>>
>>> Than I've installed lattest nvidia cuda-5.5 drivers (driver, toolkit as
>>> well as samples) and define all paths to the bash as well as libconf files.
>>>
>>> When I tried to lunch namd I've obtain error that libcudart.so.4 is not
>>> found (indeed in the cuda/lib and lib64 only libcudart.so.5 files are
>>> present).
>>>
>>> I've found libcudart.so.4 only in VMDs folder and when I've provide it
>>> in bash Namd have been worked.
>>>
>>> Should I change some modification to replace lib libcudart.so.5 to
>>> libcudart.so.4 ? My namd output is
>>>
>>> CharmLB> Load balancer assumes all CPUs are same.
>>> Charm++> Running on 1 unique compute nodes (12-way SMP).
>>> Charm++> cpu topology info is gathered in 0.001 seconds.
>>> Info: NAMD CVS-2013-10-31 for Linux-x86_64-multicore-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 60500 for multicore-linux64-iccstatic
>>> Info: Built Thu Oct 31 02:26:47 CDT 2013 by jim on lisboa.ks.uiuc.edu
>>> Info: 1 NAMD CVS-2013-10-31 Linux-x86_64-multicore-CUDA 1
>>> drunk_telecaster own
>>> Info: Running on 1 processors, 1 nodes, 1 physical nodes.
>>> Info: CPU topology information available.
>>> Info: Charm++/Converse parallel runtime startup completed at 0.00467801 s
>>> Did not find +devices i,j,k,... argument, using all
>>> Pe 0 physical rank 0 binding to CUDA device 0 on drunk_telecaster:
>>> 'GeForce GTX TITAN' Mem: 6143MB Rev: 3.5
>>> FATAL ERROR: No simulation config file specified on command line.
>>>
>>>
>>> does it means the both of my GPUs ready to use ?
>>>
>>> James
>>>
>>>
>>> 2013/10/30 Norman Geist <norman.geist_at_uni-greifswald.de>
>>>
>>>> 1 - No.
>>>>
>>>> 2 - Doesn't matter here most of the cases.
>>>>
>>>>
>>>>
>>>> Norman Geist.
>>>>
>>>>
>>>>
>>>> *Von:* owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] *Im
>>>> Auftrag von *James Starlight
>>>>
>>>> *Gesendet:* Mittwoch, 30. Oktober 2013 14:13
>>>> *An:* Namd Mailing List
>>>>
>>>> *Betreff:* Re: namd-l: Two GPU-based workstation
>>>>
>>>>
>>>>
>>>> Some extra questions-
>>>>
>>>> 1- Do I need special drivers optimizing dual GPU in Debian ?
>>>>
>>>> 2- Should I compile NAMD from sources for optimal performance ?
>>>> Previouslu I've used NAMD from Binaries ( using 1 gPU + 4 cores of i5 )
>>>>
>>>> James
>>>>
>>>>
>>>>
>>>> 2013/10/29 Norman Geist <norman.geist_at_uni-greifswald.de>
>>>>
>>>> >Just remember, NAMD is very memory bandwidth hungry.
>>>>
>>>>
>>>>
>>>> Guess you mean PCIE bandwidth hungry?
>>>>
>>>>
>>>>
>>>> Norman Geist.
>>>>
>>>>
>>>>
>>>> *Von:* owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] *Im
>>>> Auftrag von *Ajasja Ljubetic
>>>> *Gesendet:* Dienstag, 29. Oktober 2013 14:34
>>>> *An:* James Starlight
>>>> *Cc:* Norman Geist; Namd Mailing List
>>>>
>>>>
>>>> *Betreff:* Re: namd-l: Two GPU-based workstation
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> As I've told I have typical desktop with 6 pci/e slots and 1 coreI7
>>>> (4 cores) + 2 GPU + 4 RAM slots (each of 4gb or 8gb I dont remember it now
>>>> clearly :) ). I'd like to launch simulations ( water soluble as well as
>>>> membrane proteins using NAMD with the explicit solvents (50k and 80k atoms
>>>> resp) using both GPUs simultaneously and CPU for one run.
>>>>
>>>> What another extra modification of my desktop as well as simulation
>>>> parameters should I take into account? Are any other specified drivers
>>>> needed for typical Linux-based multi GPU workstation?
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Make sure to pick a motherboard with two 16x speed PCIE ports.
>>>> (Probably PCIE 3.0?). Personally I don't think you will see the scaling you
>>>> desire. I.e., the GPUs will be underutilized. But then again, YMMV. Just
>>>> remember, NAMD is very memory bandwidth hungry.
>>>>
>>>>
>>>>
>>>> Best regards,
>>>>
>>>> Ajasja
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>> --
>> Aron Broom M.Sc
>> PhD Student
>> Department of Chemistry
>> University of Waterloo
>>
>
>

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:55 CST