Problems compiling the BLACS Tester

Open discussion regarding features, bugs, issues, vendors, etc.

Problems compiling the BLACS Tester

Postby Zahid_Aziz » Wed Feb 08, 2006 8:05 am

I have been trying to compile the Blacs Tester files (from the archive www.netlib.org/blacs/blacstester.tgz) using the supplied 'make' file. Our compilers are g77 and gcc. Doing 'make' produces a large number of warning messages of the following form:

----------
blacstest.f: In subroutine `runtests':
blacstest.f:150: warning:
CALL RUNTESTS( MEM, MEMLEN, CMEM, CMEMSIZ, PREC, NPREC, OUTNUM,
1
blacstest.f:178: (continued):
SUBROUTINE RUNTESTS( MEM, MEMLEN, CMEM, CMEMLEN, PREC, NPREC,
2
Argument #1 (named `mem') of `runtests' is one type at (2) but is some other type at (1) [info -f g77 M GLOBALS]
----------

Looking at the Fortran source code of blacstest.f, this is because in the main program the array MEM is declared as Double Precision but in the subroutine it is declared as Integer. I presume there is some reason behind this difference in the declarations.

Although these appear as warnings, the Make fails with the message:

make: *** [blacstest.o] Error 1

and no object files or exectuable are created, as far as I can see.

I seem to have compiled the Blacs themselves correctly, and so my Bmake.inc should be correct.

Any help/advice would be appreciated.

Zahid Aziz
The University of Nottingham
Nottingham
England
Zahid_Aziz
 
Posts: 2
Joined: Wed Feb 08, 2006 7:18 am

Postby Julie » Wed Feb 08, 2006 8:47 am

Zahid,

change BLACS/TESTING/Makefile line 39 from:
Code: Select all
$(F77) $(F77NO_OPTFLAGS) -c $*.f
to:
Code: Select all
$(F77) $(F77NO_OPTFLAGS) -fno-globals -fno-f90 -fugly-complex -w -c $*.f

For more information, see: http://www.netlib.org/blacs/blacs_errata.html#TestErrata

It should solve your problem
Sincerly
Julie
Julie
 
Posts: 299
Joined: Wed Feb 23, 2005 12:32 am
Location: ICL, Denver. Colorado

Postby Zahid_Aziz » Thu Feb 09, 2006 7:54 am

Julie,

Thanks very much. That has worked. When I now run the Blacs Tester I receive the following error messages. These might well be related to our system, but perhaps you could give us a few pointers on what might be producing them:

------

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 ERROR 'Illegal grid (2 x 2), #procs=1'
from {-1,-1}, pnum=0, Contxt=-1, on line -1 of file 'BLACS_GRIDINIT/BLACS_GRIDMAP'.

[0] MPI Abort by user Aborting program !
[0] Aborting program!
FEP:ERROR Illegal resource specification, or
'/home/cczaziz/blas_testing/./xFbtest_MPI-LINUX-0' is not an SCore application.

------

SCore is the implementation of MPI on our HPC Grid.

Thanks,
Zahid Aziz
Zahid_Aziz
 
Posts: 2
Joined: Wed Feb 08, 2006 7:18 am

Postby Julie » Thu Feb 09, 2006 4:53 pm

Zahid,

It seems that your are running the BLACS tester on only 1 processor.
BLACS ERROR 'Illegal grid (2 x 2), #procs=1'

The BLACS tester needs at least 4 processors.

Don't pay attention to the warning message.

It hope it will solve your problem.
Sincerely
Julie
Julie
 
Posts: 299
Joined: Wed Feb 23, 2005 12:32 am
Location: ICL, Denver. Colorado


Return to User Discussion

Who is online

Users browsing this forum: Bing [Bot], Google [Bot], Yahoo [Bot] and 3 guests

cron