Dear Prof. Ross Walker,
First of all, thanks for the quick reply :) This morning I had started
over from the original files, and applied the bugfixes step by step,
but the error is still there. However, I was somewhat misinformed. I
myself have retried the same run on CPU, and the error is still there.
I have also observed that, whenever I run a simulation job instead of
minimization, the job is perfectly ok, regardless of the executable;
pmemd or pmemd.CUDA. The problem only exists for minimization
simulations.
pmemd somewhat cannot open unit 16 with the file name that I supply,
but I still cannot tackle the reason.
Later this weekend I will try to re-compile all the source code, but
I'm not too hopeful.
On Fri, Jul 13, 2012 at 4:57 PM, Ross Walker <ross.rosswalker.co.uk> wrote:
> Hi Seren,
>
> This is very strange. Are you certain that all the bugfixes applied
> correctly? Were there any error messages during the patching? - the reason
> you get fort.16 is that that is the unit used to write restart files.
> Normally a file is opened and the unit pointed to a filename. You get the
> default filename of fort.16 when the write occurs on a file that has not
> been opened. Hence the restart file is closed when it is doing the write
> which suggests to me that something is corrupt in the source code.
>
> Do you have the original AMBER 11 and AmberTools 1.5 tar files? If you do
> then you might want to try starting from a completely clean directory and
> applying the bugfixes again (both amber and AmberTools) from scratch. It
> might be quicker than trying to track down what went wrong previously.
>
> All the best
> Ross
>
>> -----Original Message-----
>> From: Seren Soner [mailto:seren.soner.gmail.com]
>> Sent: Friday, July 13, 2012 6:44 AM
>> To: AMBER Mailing List
>> Subject: [AMBER] no restart file from pmemd.cuda in Amber 11
>>
>> Dear All,
>>
>> I have installed the final version of Amber 11, with all the bugfixes
>> for both AmberTools 1.5 and Amber 11 itself. However, when I run
>> pmemd.cuda, the restart file does not exist, but instead, there is a
>> fort.16 file, very similar to the structure of restart file:
>>
>> 57554
>> 62.6224520 33.2667655 16.8053255 62.5905991 32.3885900
>> 16.2838384
>> 63.5297068 33.7019482 16.6271135 61.8798230 33.8253635
>> 16.3817990
>> 62.3808115 32.9934161 18.2527516 62.9504305 32.1037091
>> 18.5308745
>> 62.9082214 34.1630195 19.1157645 62.2458675 34.3271875
>> 19.9668928
>> 62.9104991 35.0781751 18.5228503 64.3270737 33.8930776
>> 19.6589332
>> 64.9828308 33.5857061 18.8464886 64.2848324 33.0771144
>> 20.3810183
>> 64.9278300 35.1342910 20.3401949 64.2640329 35.4574561
>> 21.1434712
>> 65.0147031 35.9314095 19.6012539 66.3180777 34.8342963
>> 20.9168274
>> 67.0006435 34.5960489 20.0970415 66.2452251 33.9541394
>> 21.5594070
>> 66.8660280 35.9637909 21.7070618 67.7669238 35.7166182
>> 22.1083976
>> ...
>>
>> On top of my output file, I also can see that the restart file address
>> is obtained from the argument:
>>
>> | RESTRT: /home/../xxx.rst
>>
>> I have compiled Amber 11 using Intel compilers, and since I already
>> have patched for bugfix.22 and bugfix.23, what could be the problem ?
>>
>> Best regards,
>> Seren Soner
>>
>> _______________________________________________
>> 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
--
Seren Soner
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Jul 13 2012 - 07:30:03 PDT
