Re: [AMBER] Failure minimizing pamr7 files from A1 tutorial with NAMD=CUDA

From: Francesco Pietra <chiendarret.gmail.com>
Date: Thu, 20 Oct 2011 22:29:26 +0200

You are right and I was naive enough not to recognize that from the
values 90 90 90 (having set an orthogonal box!)

francesco

On Thu, Oct 20, 2011 at 6:14 PM, Ismail, Mohd F. <farid.ou.edu> wrote:
> I believe the last three values is the angle between the vectors.
>
> --Farid
>
> ________________________________________
> From: Francesco Pietra [chiendarret.gmail.com]
> Sent: Thursday, October 20, 2011 5:24 AM
> To: AMBER Mailing List
> Subject: Re: [AMBER] Failure minimizing pamr7 files from A1 tutorial with       NAMD=CUDA
>
> Hello Ross:
> Just a quick (final) question: do the last line of the .inpcrd file
> for a rectangular box represent the cellbasisvectors (xyz, first three
> values) and the center of the box (xyz, last three values)?
>
>  64.0352560  68.0597930  61.0211890  90.0000000  90.0000000  90.0000000
>
> After that Ill'go asking at the NAMD desk.
>
> Thanks a lot
> francesco
>
> ***************************
> If you want to know why I am asking that:
>
> I used them in the .conf file:
> # periodic settings
> cellBasisVector1    64.0352560    0.0   0.0
> cellBasisVector2     0.0    68.0597930   0.0
> cellBasisVector3     0.0    0.0  61.0211890
> cellOrigin   90.0000000  90.0000000  90.0000000
>
> I did that because the .prmtop/.inpcrd of the A1 tutorial saved aafter
> "solvatebox 1PLC TIP3PBOX 12" let to perform a correct minimization
> with NAMD-CUDA multiple GTX-580 (juust 1000 steps; I forgot to check
> the gradient), while they do not open correctly in VMD (just a
> straigth segment instead of a box, so that I could not measure the
> box). The .prmtop/.dcd after the minimization open correctly in VMD as
> a box with the protein at the center.
>
> After saving .prmtop/.inpcrd from leap I also saved
>
> "savepdb 1PLC 1PLC.box.pdb"
>
> but this does not help VMD.
> Incidentally, as I was curious about this matter, I simply copied to
> ambertools1.5 xleap and xaLeap from amber10. It works, except for some
> fonts not converted and +Icon pixmap "xleap_icon" not found.
> ************************
>
>
>
>
>
>
> On Thu, Oct 20, 2011 at 10:17 AM, Ross Walker <ross.rosswalker.co.uk> wrote:
>> Hi Francesco,
>>
>>> The Giambasu-Case directions - albeit restricted to cpu only - seemed
>>> to invite to talk with amber, too. Sorry for the intrusion following
>>> my misunderstanding.
>>
>> It's not an intrusion. I think you might have more chance of success on the NAMD forums. Especially if your topology and coordinate files match and work in VMD and the CPU version of NAMD. Then it might be something subtle going wrong inside the code and you might need an expert in the code to look at it.
>>
>>> I'll do. I assumed that the files provided with the tutorial are
>>> correct. On the other hand, they load correctly to VMD. Better, I'll
>>
>> Try just running on a single processor/single GPU - it might be an issue with multi GPU NAMD. Also see if VMD offers you any error messages in the console box when loading the files. I think the key will be to see if it works in the cpu version. You can download a binary of namd and just quickly try it with the single cpu version and see if that works. That will quickly tell you if it is the input files or a bug in the code.
>>
>>> run myself xleap for a rectangular box. Curiously, xleap is not in my
>>> installation of amber11/ambertools1.5, while it is in the installation
>>> of amber10.
>>
>> This sounds like it got skipped during installation because an X11 development library was missing. Did you compile the amber10 version on a different machine? - You could try running '$AMBERHOME/AmberTools/src/configure gnu'   and see if it mentions anything about skipping the building of xleap. Normally this is because xorg-X11-devel (or similar) is not installed.
>>
>>> > On another note why not just use GPU AMBER  (PMEMD)?
>>>
>>> Present status of the Country, with drastic budgetary cuts, have
>>> forced us to priorities. I still am at amber10/ambertools1.5. More
>>> importantly, what i want to simulate can't be done with current
>>> amber11.
>>
>> Fee waivers are available in circumstances where people do not have the funds to purchase a license. You can make a request to amber-admin.biomaps.rutgers.edu
>>
>>
>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Oct 20 2011 - 13:30:03 PDT
Custom Search