The ABINIT Users Mailing List ( CLOSED )

[james raynolds <jamesraynolds@yahoo.com>]

To all,

The IBM compiler problem reported yesterday has been
solved.  The solution is just as confusing as the
problem.  The consultant solved the problem with a few
changes to the compiler options.

The transcript of her message follows:

Sincerely,
Jim Raynolds

-----------------------message from consultant
--------------

As you I have successfully done the following:

make allseq

I copied makefile_macros to makefile_macros.bak The
changes I made are
as follows:

diff makefile_macros makefile_macros.bak
21c21
< FFLAGS=-O2 -qnolm -qarch=pwr3 -qstrict -g -q32 -d
---
> FFLAGS=-O3 -qnolm -qarch=auto -qstrict -g -q32 -d -C
25c25
< FFLAGS_LIBS=-O2 -qnolm -qarch=pwr3 -qstrict -qfixed
-g -q32 -d
---
> FFLAGS_LIBS=-O3 -qnolm -qarch=auto -qstrict -qfixed
-g -q32 -d -C

I assume you are plannng to execute the code on the
batch nodes, hence
the -qarch=pwr3  -O3 requires too much time and memory
to compile on
squall interactive node which is an IBM F50. The batch
nodes are IBM
9076. The -C option can slow your execution down and
since you 
expressed
concern about this I removed it. I think you will that
the above 
changes
will produce the results you are looking for.  -MAC



                
__________________________________ 
Do you Yahoo!? 
Yahoo! Small Business - Try our new resources site!
http://smallbusiness.yahoo.com/resources/