public inbox for gcc-help@gcc.gnu.org
 help / color / mirror / Atom feed
* Analysis results differ with different versions of g77 compilers?
@ 2004-03-10  4:01 Lu, Yuan
  2004-03-10 14:46 ` Eljay Love-Jensen
  0 siblings, 1 reply; 2+ messages in thread
From: Lu, Yuan @ 2004-03-10  4:01 UTC (permalink / raw)
  To: 'gcc-help@gcc.gnu.org'

Hi there,

I have a strange problem: When I copied the Fortran programs 
(compiled with GNU F77 2.96 20000731) from a
Red Hat 7.3 system to a Red Hat 9.0 system, and
run the programs, I got one results set.

After I recompiled the same Fortran programs with gcc 3.2.2 20030222
on the same Redhat 9.0 system , and ran the same programs with same 
input data set as above, I got a different results set. Usually the 
results differs from the 4-th decimal point. 

Any body knows what's the reasons why I got different results? 

Many thanks.

Yuan

^ permalink raw reply	[flat|nested] 2+ messages in thread
* Re: Analysis results differ with different versions of g77 compilers?
  2004-03-10  4:01 Analysis results differ with different versions of g77 compilers? Lu, Yuan
@ 2004-03-10 14:46 ` Eljay Love-Jensen
  0 siblings, 0 replies; 2+ messages in thread
From: Eljay Love-Jensen @ 2004-03-10 14:46 UTC (permalink / raw)
  To: Lu, Yuan, 'gcc-help@gcc.gnu.org'

Hi Yuan,

I'm not 100% sure, but my guess would be that one or both programs were not 
compiled with strict IEEE 754 compliance enabled.

Without strict compliance, the compiler is at liberty to take advantage of 
optimizations which may result in twitchy differences in least significant 
digits of floating point numbers.

--Eljay

^ permalink raw reply	[flat|nested] 2+ messages in thread
end of thread, other threads:[~2004-03-10 12:33 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-03-10  4:01 Analysis results differ with different versions of g77 compilers? Lu, Yuan
2004-03-10 14:46 ` Eljay Love-Jensen

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).