[gmx-users] Reducing Standard Error output

[Mark Abraham]

Jack Shultz wrote:
> Even though I can sucessfully run this when I remove the verbose flag,
> it causes another problem. We keep trac of the progress of indivual
> simulations using a code modfication in sim_util.c
> 
> 124:      f = fopen("progress.txt", "w");
> 125-      if (!f) return;
> 126-      fprintf(f, "%g", (step - ir->init_step) / (float) ir->nsteps);
> 127-      fclose(f);
> 
> Apparently this code is only used if the app is in verbose mode. So I
> guess we have two choices:
> 1) either find the cause of the lincs warnings

That is always necessary if you don't want an expensive random number 
generator. Your simulation system is ill conditioned, and the lincs 
warnings will exist and be reported in at least the .log file regardless 
of the verbosity.

> 2) modify the code so it reports progress regardless of verbose flag

Sure, but that's independent of the former. You should consider other 
modifications too - opening and closing a file every integration step 
might be prohibitively expensive. Since from previous threads you're 
monitoring this progress file with some other process to report to the 
user, you should consider more efficient means of such communication, 
such as an inter-process socket.

Mark

> On Thu, Dec 3, 2009 at 7:39 PM, Jack Shultz <js at drugdiscoveryathome.com> wrote:
>> yes the environment I was running in was the boinc wrapper and its
>> supposed to somehow return the standard error output. It probably got
>> overloaded. It is curious why I only had the problem on windows 64 but
>> I'm not going to worry about it.
>>
>> On Thu, Dec 3, 2009 at 6:33 PM, Mark Abraham <Mark.Abraham at anu.edu.au> wrote:
>>> Jack Shultz wrote:
>>>> Thanks,
>>>>
>>>> I just figured out removing the -v will reduce output. Interestingly I
>>>> only have this problem with 64-bit Windows hosts. I have not observed
>>>> it in any others and I have a fairly diverse environment on this
>>>> project. For now I can say removing the -v gets the app to finish the
>>>> run.
>>> mdrun -v will have no effect on correct completion, unless the use of -v is
>>> causing some I/O buffer external to GROMACS to over-run. I suspect your
>>> observation indicates you were comparing apples and oranges.
>>>
>>> Mark
>>> --
>>> gmx-users mailing list    gmx-users at gromacs.org
>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>> Please search the archive at http://www.gromacs.org/search before posting!
>>> Please don't post (un)subscribe requests to the list. Use the www interface
>>> or send it to gmx-users-request at gromacs.org.
>>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>>
>>
>>
>> --
>> Jack
>>
>> http://drugdiscoveryathome.com
>> http://hydrogenathome.org
>>
> 
> 
>