[gvd-users] My first session on a supercomputer

[Arjan van Dijk]

>Have you tried running it in a single-processor environment (no -mp)?

Yes, the program was already compiled for single processor.



>Yes, gvd works, but it's not a debugger - only a front-end.

Of course. I think that I can assume that gvd itself works, but that 
the debugger does not want to talk with my executable.


>Perhaps you use g77? Or a different combination of compiler, debugger
>and compiler options?

Probably. I just managed to get the example program, with just a few 
lines, running in gvd on the supercomputer. I had to use the g77 
compiler instead of the f90 compiler. The problem is that my climate 
model is pure fortran90 and the g77 version on the supercomputer does 
not eat f90. The command "g77 --version" gives:



>GNU Fortran 0.5.26 20020220 (release)
>Copyright (C) 2001 Free Software Foundation, Inc.
>For more version information on components of the GNU Fortran
>compilation system, especially useful when reporting bugs,
>type the command `g77 --verbose'.
>
>GNU Fortran comes with NO WARRANTY, to the extent permitted by law.
>You may redistribute copies of GNU Fortran
>under the terms of the GNU General Public License.
>For more information about these matters, see the file named COPYING
>or type the command `info -f g77 Copying'.


The man pages make clear that files of type *.f90 are not recognized 
and there is no reference to f90 anyhow. Would it be fair to ask the 
systems support of the supercomer to install a more recent g77?


>  > Would it be a problem if I used a non-gnu compiler to make
>>  my executable from the Fortran sources? I used the MIPSpro 7 Fortran
>>  90 compiler, which is the default compiler that is activated by
>>  "f90".
>
>That would be very much a problem if gdb does not support the
>environment you run in. Is it an older version of gdb (current is 5.x)?
>Is it 64bit compatible (I guess your software is compilled for 64 bit)?
>etc.



I recompiled the short test program with the f90 compiler and tried 
to switch to dbx, which by the way does exist on the supercomputer. 
Switching to dbx was not allowed, since the tick-mark in the 
change-compiler menu was grey and inactive. Entering the name dbx in 
the lower field for debuggernames in the change-debugger menu did not 
help either: "unexpected internal error; please report" was shown at 
the bottom of gvd (by the way, in dutch "gvd" is a strong curse...).

Strange enough the man pages of the f90 compiler do not tell for 
which debugger the debug-info, that is generated by the -g compile 
option, is intended. When I tried to run gdb on the climate model, 
there was an error message with some text about Dwarf (see my first 
message). Wasn't Dwarf a debugger? Why would gdb complain about 
Dwarf? The gdb version is:


GNU gdb 5.2
Copyright 2002 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "mips-sgi-irix6.5".


This is relatively recent, so I would hope that it can be used.

I now see two options to proceed:

-Try to install a newer g77 version that eats f90. I don't like this 
option very much, since the whole climate model is set up to use all 
sorts of optimizations and compile-switches that are specific for 
this compiler.

-Use a different debugger or make the debug-info readable for gdb, but how?

Who sees better options or can help otherwise?

Thanks,

Arjan


-- 
Arjan van Dijk
Institute for Marine and Atmospheric Research Utrecht
Faculty of Physics and Astronomy
Utrecht University
Princetonplein 5
NL - 3584 CC  Utrecht
The Netherlands

phone: +31 30 2532815
fax:   +31 30 2543163
e-mail: mailto:A.vanDijk@phys.uu.nl
homepage: http://www.phys.uu.nl/~vndijk