getting compilation listing

[Tim Prince]

dozsagyorgy@swissinfo.org wrote:
> Dear Sir/Madam,
> I am running Fortran77 programs on Fujitsu-Siemens PC using MinGW under Windows
> XP. 
> Do you know a compiler option to get a compilation listing from g77 onto
> a disc file? 
> My second problem is that I am not getting the same accuracy I used to get
> on the SUN system under Solaris operating system. Is it down to MinGW or
> Fujitsu-Siemens arithmetic unit or Windows XP?
This is somewhat off topic here, since you don't ask directly about 
gfortran.
If you mean what I think, g77 and gfortran don't offer a compilation 
listing, nor do many
current compilers.  For g77, the gcc-help list would be more on target.
'info g77' or 'info gfortran,' or equivalent documentation at 
gcc.gnu.org, are the first line of inquiry.
You would have to ask a better posed question for the second one. We 
can't guess what types of CPU you are comparing or what options you 
chose.  With options -march=nocona -mfpmath=sse for a Xeon or Opteron, 
one would expect numerical results similar to those from a Sun SPARC 
system. If you used g77 or gfortran with the same options on a Sun 
Opteron system and a Windows system, results should be much the same.  I 
don't know any case where Fujitsu or Siemens would have an arithmetic 
unit which would give different results from another one used with the 
same CPU family.  As Fujitsu have sold both SPARC and Xeon systems, as 
well as others, it's not totally clear what you mean.  If you used a 
Xeon or Opteron CPU without the sse option, you would normally expect 
more accurate results than you would get from the SPARC.  Many people 
have considered that objectionable.  It is typical to find discrepancies 
with extended precision, as it is unlikely to be used consistently.
If you are looking for identical numerical answers with different 
compilers, that is unlikely except for the simplest cases.
Appropriate, more precise, statement of the question might make it 
suitable for comp.lang.fortran.