VMD-L Mailing List

From: Ajasja Ljubetič (ajasja.ljubetic_at_gmail.com)
Date: Tue Mar 01 2016 - 05:02:32 CST

Hi Davit!

For the conversion you could try the 64 bit catDCD that Axel cross-compiled
a while ago for.
http://rpm.lammps.org/windows/testing/

Best,
Ajasja

On 1 March 2016 at 11:37, Davit Hakobyan <davhak_at_gmail.com> wrote:

> Thanks very much for your replies.
>
> The over 5GB RAM memory is left free when this problems occurs in VMD so
> most likely the problem relates to 4GB limit which Josh clarified. The
> problem is that in this particular case I use VMD to convert the XTC file
> to DCD format and I really need all the frames. So if there is no way to
> convince VMD to read all 2000 frames (with stride 1) then I need to think
> of some workaround like splitting the whole XTC to portions and feed them
> separately to VMD and to have several sequential DCD files in the output.
>
> Please let me know if a better solution exist.
>
> Again, thanks a lot for any suggestion.
>
>
> On Mon, Feb 29, 2016 at 10:35 PM, Josh Vermaas <vermaas2_at_illinois.edu>
> wrote:
>
>> Hi Davit,
>>
>> This sounds like you are on Windows and you have just passed the memory
>> addressing limit vmd inherits by being a 32bit executable, which is 4gb.
>> Once enough trajectory is loaded so that it passes the threshold, vmd can't
>> address a memory location and crashes. Striding to reduce the loaded size
>> is normally how people get around this, and use bigdcd ideas for analysis.
>> If you go this route, I'd recommend converting the xtc to something else,
>> since xtc frames aren't the same size on disk and so vmd can't read
>> arbitrary frames efficiently, since it must read every intermediate frame.
>> Josh
>> On Feb 29, 2016 11:28 AM, Davit Hakobyan <davhak_at_gmail.com> wrote:
>>
>> Dear All,
>>
>> By now I was successfully loading XTC trajectory file containing 1500
>> frames to VMD. Now as the simulation is still running and additional 500 ns
>> were generated When trying to load the updated trajectory file containing
>> 2000 frames VMD either crashes after loading some 1600 frames (if all the
>> atoms are visible) or just stops randomly after loading 1623 ((the number
>> can be different at different load attempts) frames (this is when only a
>> selection like "resid 1" is active).
>>
>> Why I am basically sure that XTC file is correct is because when I try to
>> load the same trajectory file with Stride 10, VMD then loads exactly 200
>> frames.
>>
>> The XTC is generated on 64 bit Linux system and VMD is running on 64 bit
>> Windows 7 with 12 GB of RAM, so memory is not a problem.
>>
>> The system size is ~74000 atoms.
>>
>> Does someone have any idea how to resolve this?
>>
>> Thanks a lot for any help.
>>
>>
>