Just to follow up for the web archives -- as was suggested in several
of the other, related, threads today, this is not the right thing to do
(the mpif.h that you linked to is part of the LAM distribution, but it
is actually part of ROMIO, and only for SGI's MPI -- not LAM/MPI).
You do *not* want an mpif.h in the BLACS directory -- do not make a sym
link to do so.
Trust mpif77 to find the right mpif.h for you -- it puts the Right
command line arguments to find its own mpif.h.
From your description, it looks like /usr/include/mpif.h belongs to
LAM/MPI, but as people pointed out:
- you can view /usr/include/mpif.h and read the comments at the top --
if it belongs to LAM/MPI, it will be quite obvious (it says "This file
is part of the LAM/MPI software package")
- I'm guessing that you had LAM installed via RPM. You can query to
see if /usr/include/mpif.h belongs to the same RPM as /usr/bin/mpicc
with the following commands:
rpm -qf /usr/include/mpif.h
rpm -qf /usr/bin/mpicc
If they're from the same RPM, they should both give the same answer.
- But what might be easiest is to uninstall this LAM installation (if
it was an RPM, then perhaps "rpm -e" will uninstall it) and install a
fresh, known copy that puts you at a simple, known state where you're
absoultely sure that mpif77 matches mpif.h, etc.
On Mar 30, 2005, at 10:52 AM, Srinivasa Prade Patri wrote:
> HI!
> Thanks for your suggestions.Iam still not able to install BLACS
> library after changing the include file mpif.h.
>
> With command find -name mpif.h , i can see there are two locations
> for mpif.h
>
> ./usr/include/mpif.h
> ./usr/src/caos/SOURCES/lam-7.1.1/romio/adio/sgi/mpi3.1/mpif.h
> ./home/BLACS/INSTALL/mpif.h
> ./home/BLACS/TESTING/mpif.h
>
> and with mpif77 -showme command ,i see
>
> g77 -I/usr/include -pthread -L/usr/lib64 -llammpio -llamf77mpi -lmpi
> -llam -lutil -ldl
>
> Hence i used /usr/include/mpif.h for installing BLACS library
> and not able to run the test suite
>
> I even tried with include file at
> /usr/src/caos/SOURCES/lam-7.1.1/romio/adio/sgi/mpi3.1. and iam getting
> the following error during make mpi
> -----------------------------------------------------------------------
> -------------
> ln -s /usr/src/caos/SOURCES/lam-1.1/romio/adio/sgi/mpi3.1/mpif.h
> mpif.h
> mpif77 -c -Wno-globals -O bi_f77_mpi_attr_get.f
> mpif.h: In subroutine `bi_f77_mpi_attr_get':
> mpif.h:294:
> pointer (MPI_BOTTOM_PTR, MPI_BOTTOM)
> 1 2
> Unrecognized statement name at (1) and invalid form for assignment or
> statement-function definition at (2)
> mpif77: No such file or directory
> make[2]: *** [bi_f77_mpi_attr_get.o] Error 1
> make[2]: Leaving directory `/home/BLACS/SRC/MPI/INTERNAL'
> make[1]: *** [INTERN] Error 2
> make[1]: Leaving directory `/home/BLACS/SRC/MPI'
> make: *** [MPI] Error 2
> -----------------------------------------------------------------------
> ----
>
> Since Iam able to compile with usr/include/mpif.h, Iam
> confused with which mpif.h i should go. I would greatly appreciate
> your suggestions.
>
> Thanking you.
>
> Regards
> Srinivasa Patri
>
> -----Original Message-----
> From: Hugh Merz <merz_at_[hidden]>
> To: sppatr2_at_[hidden], General LAM/MPI mailing list <lam_at_[hidden]>
> Date: Wed, 30 Mar 2005 07:20:33 -0500 (EST)
> Subject: Re: LAM: Problem running the tester for BLACS
>
> This is almost certainly a mix-up between mpich and lam files, as
> indicated in the archives here:
>
> http://www.lam-mpi.org/MailArchives/lam/2001/09/3306.php
>
> Are you sure you have the details of your mpi installation correct? In
> your Bmake.inc:
>
> --snip--
> MPIdir = /usr
> MPILIBdir = $(MPIdir)/lib64
> MPIINCdir = /usr/include
> MPILIB = $(MPILIBdir)/libmpi.a
> --/snip--
>
> My guess is that there is an mpich mpif.h in /usr/include - you should
> change the above so that it finds you LAM installation.
>
> Hugh
>
>
> On Wed, 30 Mar 2005, Srinivasa Prade Patri wrote:
>
>> Hi!
>> I apologize for not being clear in my last mail.BLACS is still not
>> working and the following are the macro settings i have used...
>>
>> SYSINC =
>> /* left blank*/
>>
>> INTFACE = -Df77IsF2C
>> /* when i ran xintface i got the following output
>> For this platform, set INTFACE = -Df77IsF2C */
>>
>> TRANSCOMM = -DUseMpi2
>> /* when i ran xtc_CsameF77 i got the following output
>> [patri_at_kemc ~]$ mpirun -v -np 2 /home/BLACS/INSTALL/EXE/xtc_CsameF77
>> 23986 /home/BLACS/INSTALL/EXE/xtc_CsameF77 running on n0 (o)
>> 23987 /home/BLACS/INSTALL/EXE/xtc_CsameF77 running on n0 (o)
>> If this routine does not complete successfully,
>> Do _NOT_ set TRANSCOMM = -DCSameF77
>>
>>
>> Do _NOT_ set TRANSCOMM = -DCSameF77 */
>>
>> F77 = mpif77
>>
>> CC = mpicc
>>
>> I dont know exactly where iam doing wrong. Iam attaching the
>> Bmake.inc with this mail.
>>
>> The following are the errors When iam executing the test run with
>>
>> [patri_at_e01 ~]$ mpirun -v -np 2
>> /home/BLACS/TESTING/EXE/xFbtest_MPI-LINUX-0
>>
>> 1477 /home/BLACS/TESTING/EXE/xFbtest_MPI-LINUX-0 running on n0 (o)
>> 1199 /home/BLACS/TESTING/EXE/xFbtest_MPI-LINUX-0 running on n1
>> BLACS WARNING 'No need to set message ID range due to MPI
>> communicator.'
>> from {-1,-1}, pnum=0, Contxt=-1, on line 18 of file 'blacs_set_.c'.
>>
>> BLACS WARNING 'No need to set message ID range due to MPI
>> communicator.'
>> from {-1,-1}, pnum=1, Contxt=-1, on line 18 of file 'blacs_set_.c'.
>>
>> MPI_Comm_group: invalid communicator: Invalid argument (rank 1,
>> MPI_COMM_WORLD)
>> Rank (1, MPI_COMM_WORLD): Call stack within LAM:
>> Rank (1, MPI_COMM_WORLD): - MPI_Comm_group()
>> Rank (1, MPI_COMM_WORLD): - main()
>> MPI_Comm_group: invalid communicator: Invalid argument (rank 0,
>> MPI_COMM_WORLD)
>> MPI_Comm_group: invalid communicator: Invalid argument (rank 0,
>> MPI_COMM_WORLD)
>> ----------------------------------------------------------------------
>> -------
>> One of the processes started by mpirun has exited with a nonzero exit
>> code. This typically indicates that the process finished in error.
>> If your process did not finish in error, be sure to include a "return
>> 0" or "exit(0)" in your C code before exiting the application.
>>
>> PID 1477 failed on node n0 (10.5.0.1) with exit status 1.
>> ----------------------------------------------------------------------
>> -------
>>
>> Thanking you.
>>
>> Regards
>> Srinivasa Patri
>>
>>
>>
>
>
>
>
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
{+} Jeff Squyres
{+} jsquyres_at_[hidden]
{+} http://www.lam-mpi.org/
|
