I have compiled OMPI with g95. Later, sander.MPI was compiled with
mpif90 (OPMI/g95...). I met few warnings like I described before, but
sander (g95, gcc 3.2) and sander.MPI (mpif90/95, gcc 3.2) were running
(but very slow - different topic). However, compiling PMEMD I get
significant amount (if it helps I can post those). So, having in mind
that previous compilation of sander.MPI runs, but slow I thought MAYBE
it is important. As far i have very limited understanding of Fortran/MPI
programing, so sorry guys for sometime stupid questions.
One more thing. In general the cluster I use is little maintained, so
all installations I do locally. Admin of our cluster installed gcc 4.4.x
(c,c++). I suspect also that I need to recompile OMPI ??? Because right
now I'm not sure what headers and libraries will be used during compilation.
Arturas
Ross Walker wrote:
> Hi Arturas,
>
> I see nothing wrong with this code, it is definitely allocated as an integer
> in the first case and is being used in the bcast statement as mpi_integer -
> so my guess would be that your mpi installation was not compiled with the
> same compiler or compiler options as you are using to compile PMEMD. I.e.
> perhaps the mpi installation was somehow built with default 64 bit integers
> and amber is being build with 32 bit integers. This is just a guess though.
> I would check what "mpif90 -show" returns - does it return the same compiler
> as you are using for AMBER?
>
> Another option would be to edit the config.h file for pmemd and replace the
> fortran compiler with mpif90 and see if that helps - then at least it will
> use the same compiler as was used for mpi.
>
> I suspect that while you get an executable it will segfault as soon as you
> run it. Although it may just be the compiler spitting out random warnings
> for some reason. Run the test cases in parallel and see what happens.
>
> All the best
> Ross
>
>
>> -----Original Message-----
>> From: owner-amber.scripps.edu [mailto:owner-amber.scripps.edu] On Behalf
>> Of Arturas Ziemys
>> Sent: Wednesday, July 02, 2008 3:11 PM
>> To: amber.scripps.edu
>> Subject: AMBER: Inconsistent types (INTEGER(4)/REAL(8)) - Is it serious ?
>>
>> Hi,
>>
>> During compilation of Sander and PMED I've got warnings on basis of type
>> casting (like example below) of function arguments. Is it serious ?
>>
>> ============================================================
>> In file nmr_calls.f90:285
>>
>> call mpi_bcast(nmr_iwork, intreq, mpi_integer, &
>> 1
>> In file nmr_calls.f90:273
>>
>> call mpi_bcast(rwell, nmr_dat_dbl_cnt, mpi_double_precision, 0, &
>> 2
>> Warning (155): Inconsistent types (INTEGER(4)/REAL(8)) in actual
>> argument lists at (1) and (2)
>>
>> ============================================================
>>
>> Arturas
>>
>> -----------------------------------------------------------------------
>> The AMBER Mail Reflector
>> To post, send mail to amber.scripps.edu
>> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
>> to majordomo.scripps.edu
>>
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
> to majordomo.scripps.edu
>
>
--
Arturas Ziemys, PhD
School of Health Information Sciences
University of Texas Health Science Center at Houston
7000 Fannin, Suit 880
Houston, TX 77030
Phone: (713) 500-3975
Fax: (713) 500-3929
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
to majordomo.scripps.edu
Received on Sun Jul 06 2008 - 06:07:20 PDT
