[gmx-developers] gromacs on opensolaris - compiles but does not really run

[Mark Abraham]

Karsten Meyer wrote:
> -------- Original-Nachricht --------
>> Datum: Mon, 31 Aug 2009 12:05:50 +1000
>> Von: Mark Abraham <Mark.Abraham at anu.edu.au>
>> An: Discussion list for GROMACS development <gmx-developers at gromacs.org>
>> Betreff: Re: [gmx-developers] gromacs on opensolaris - compiles but does not	really run
> 
>> Karsten Meyer wrote:
> [...]
> 
>>> -------------------------------------------------------
>>> Program mdrun, VERSION 4.0.5
>>> Source code file: force.c, line: 1057
>>>
>>> Fatal error:
>>> Only triclinic boxes with the first vector parallel to the x-axis and
>> the second vector in the xy-plane are supported.
>>> -------------------------------------------------------
>> The most likely cause for this problem is an incorrectly formatted 
>> coordinate file supplied to grompp. Perhaps you need to find a more 
>> reliable source of inputs, since a successful test will require 
>> well-conditioned input. Possibilities include the GROMACS test set, 
>> GROMACS benchmark set or some tutorial material. Google is your friend.
>>
>> Mark
>>
> 
> Hi Mark,
> 
> it's not only the demo stuff the comes with gromacs' source tree, some other "input files" also produce this error. Does gromacs need special GNU stuff, for example flex that in case of missing generates such effects ?

No, GROMACS does not have any requirements that are not detected by 
configure.

Your first email demonstrated that various input parameters which should 
have been identical to zero were merely about 1e-39. This suggests the 
possibility of a mismatch in binary floating-point format, however other 
non-zero floats were correct. Such a mismatch could conceivably cause 
the error your report. Are you running grompp on the same architecture 
linked in the same manner as mdrun? Is the fact that you have 64-bit and 
32-bit architectures available somehow contributing?

Mark