[Pw_forum] Problem with QE-5.2.1 in building exx routines

[Axel Kohlmeyer]

On Tue, Feb 9, 2016 at 10:28 AM, Mahmoud Payami Shabestari
<mpayami at aeoi.org.ir> wrote:
> I went through all the "exx_*.f90" routines and it seems the problem was not
> what I mentioned in my last post. However, whatever the cause of the error
> is, it is in these routines (I do not know the exact reason). Maybe there is
> some small programming error which is ignored by newer versions of
> compilers?!

if you get an "internal compiler error", the problem is with the
specific compiler version.
please note that intel 11.1 is six years old and its compliance with
newer fortran standards is sub-par and their implementation has bugs.
upgrading to a newer version of the compiler is the way to proceed, or
you'll have to stick with older versions of QE or you have to place a
strategic if statement to cause the code to abort and thus disable
this feature and remove the offending code. mind you, the last of
those approaches still leaves plenty of doubts about the compiler, as
there may be other cases, where you do not get an internal compiler
error, but just miscompiled code.

in short, trying to fix code to accommodate an outdated compiler is a
fool's errand.

axel.


>
>
> -----Original Message-----
> From: Lorenzo Paulatto <lorenzo.paulatto at impmc.upmc.fr>
> To: PWSCF Forum <pw_forum at pwscf.org>
> Date: Mon, 8 Feb 2016 13:23:30 +0100
> Subject: Re: [Pw_forum] Problem with QE-5.2.1 in building exx routines
>
>
> On 8 Feb 2016 11:41 a.m., "Mahmoud Payami Shabestari" <mpayami at aeoi.org.ir>
> wrote:
>>
>> Ciao Paolo,
>> Thank you very much for your reply.
>> As it is mentioned in the UG-5.3.0, "Also, compilers that do not support
>> intrinsic calls flush, get_environment_variable,get_command_argument,
>> command_argument_count are no longer supported since v.5.2.1.", the old
>> compiler complains about "internal error" which most probably is due to the
>> above-mentioned calls present in "exx_X.f90" routines in the CP package.
> This is very unlikely, there are no calls to those intrinsic functions in
> EXX, and even if there were, they wouldn't cause an internal compiler error.
> Upgrade your compiler or ask your system administrator if you cannot.
> For others there is no problem.
>> Thank you again,
>> mahmoud
>>
>>>
>>>
>>> Hi Mahmoud, it depends upon the kind of problem. If it is an "internal
>>> compiler error", there is little that can be done. What does the old
>>> compiler complain about exactly? Paolo
>>> Dear QE users and developers,
>>>
>>> Kindly, I encountered with some problems in compiling the CP package
>>> using Intel ifort-11.1.080. The problem is referred to some line in
>>> exx_psi.f90.
>>> However, using a newer version of Intel Compiler such as Compxe-2013, the
>>> compilation proceeds without any problem.
>>> Is there any simple solution other than moving to new versions of
>>> compilers?
>>> Any suggestions is highly appreciated.
>>>
>>> Mahmoud Payami
>>> Theoretical and Computational Physics Group
>>> AEOI, Tehran, Iran
>>>
>>>
>>> _______________________________________________
>>> Pw_forum mailing list
>>> Pw_forum at pwscf.org
>>> http://pwscf.org/mailman/listinfo/pw_forum
>>
>>
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://pwscf.org/mailman/listinfo/pw_forum
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum



-- 
Dr. Axel Kohlmeyer  akohlmey at gmail.com  http://goo.gl/1wk0
College of Science & Technology, Temple University, Philadelphia PA, USA
International Centre for Theoretical Physics, Trieste. Italy.