VMD-L Mailing List

From: Peter Freddolino (petefred_at_umich.edu)
Date: Wed Aug 02 2017 - 22:11:31 CDT

Thanks Sebastien. I will bring this up off-list with the most recent
maintainer of cionize -- I suspect that a compilation flag is not being
set properly during the build.
Thanks,
Peter

On 08/02/2017 03:58 PM, Sebastian Reiter wrote:
> Dear Peter,
>
> I am using version 1.9.3 of vmd, specifically the binary distribution
> LINUX_64 OpenGL, CUDA, OptiX, OSPRay (Linux (RHEL 6.7 and later)
> 64-bit Intel/AMD x86_64 SSE, with CUDA 8.x, OptiX, OSPRay). I
> installed it by running configure with the standard options given
> below, followed by make install.
>
> LINUXAMD64 OPENGL OPENGLPBUFFER FLTK TK ACTC CUDA IMD LIBSBALL
> XINERAMA XINPUT LIBOPTIX LIBOSPRAY LIBTACHYON VRPN NETCDF COLVARS TCL
> PYTHON PTHREADS NUMPY SILENT ICC
>
> Also, the startup message of vmd indicates that multithreading is
> available:
>
> > Info) Multithreading available, 4 CPUs detected.
> > Info) CPU features: SSE2 AVX AVX2 FMA
>
> Best regards,
> Sebastian
>
>
> Am 02.08.2017 um 17:35 schrieb Peter Freddolino:
>> Dear Sebastien,
>> Given where that message is occurring, I suspect that your vmd may
>> have been built without multithreading support. Could you confirm
>> what version you are running and how you installed it?
>> Thanks,
>> Peter
>>
>>
>> On 08/02/2017 07:00 AM, Sebastian Reiter wrote:
>>> Dear Peter,
>>>
>>> thank you very much for your quick and helpful reply. You are right: I
>>> did not load a psf file, so the charges were missing. Loading the psf
>>> fixed the error and the ion positions look reasonable now.
>>>
>>> In the output, the input parameters are reported correctly at the very
>>> beginning:
>>>
>>> > Running cionize on input file cionize-temp.pqr
>>> > Ion-solute distance: 6.000000
>>> > Ion-Ion distance: 10.000000
>>> > Grid spacing: 0.500000
>>> > Boundary size: 10.000000
>>> > Max. Processors: 4
>>>
>>> However, when the actual calculation starts, it does not seem to be
>>> running in parallel, which I could also confirm by looking at the
>>> system
>>> load:
>>>
>>> > Calculating grid energies...
>>> > Multilevel summation for electrostatic potential map
>>> > using 1 processors
>>>
>>> The same happens, when double precision is used instead of multilevel
>>> summation.
>>>
>>> Best regards,
>>> Sebastian
>>>
>>> Am 02.08.2017 um 05:01 schrieb Peter Freddolino:
>>>> Dear Sebastian,
>>>> Can you confirm how you generated the structure you loaded into vmd?
>>>> In particular, are the charges set properly? I suspect that they are
>>>> not loaded given the output that you indicate that you have; for
>>>> example, you may have forgotten to load a psf file.
>>>>
>>>> Could you post the output from your job to clarify at which point the
>>>> use of only one processor was reported?
>>>>
>>>> Best,
>>>> Peter
>>>>
>>>>
>>>> On 08/01/2017 03:57 PM, Sebastian Reiter wrote:
>>>>> Dear all,
>>>>>
>>>>> I am trying to use the cionize plugin (version 1.0) in vmd 1.9.3 to
>>>>> neutralize the charges of an RNA backbone. So far, I have found
>>>>> out how
>>>>> to start the plugin and the following exemplary command does not
>>>>> yield
>>>>> an error message:
>>>>>
>>>>> cionize -mol 0 -mg -np 4 -go potential.dat -ions "{SOD 85 1}"
>>>>>
>>>>> However, the output shows that only one processor was used despite
>>>>> the
>>>>> option "-np 4" and, more importantly, the generated ions are just
>>>>> placed
>>>>> cubically around the RNA structure. A quick look into the output file
>>>>> "potential.dat" shows that cionize sets the potential to zero at
>>>>> every
>>>>> grid point.
>>>>>
>>>>> I would greatly appreciate your advice on how to use cionize
>>>>> correctly,
>>>>> so that I can make use of the parallel implementation and the ions
>>>>> are
>>>>> placed in sensible positions.
>>>>>
>>>>> Best regards,
>>>>> Sebastian
>>>>>
>>
>