[ydl-gen] gfortran under yellowdog

[Stephen Harker]

On Thu, Jul 24, 2008 at 08:45:08PM -0500, Ron Rogers Jr. wrote:
> On Wed, 16 Jul 2008 09:51:45 -0400
> jsauer at gfdi.fsu.edu wrote:
> 
> > Has anyone tried compiling fortran source using gfortran
> > under yellowdog? When compiling
> > a simple ello world program to test gfortran I recieve error
> > messages about problems with
> > libgfortranbegin.a being unreadable or not found.  Any
> > suggestions?

If you are building your own I would suggest you avoid gcc4.2x.  If
you look at the ATLAS errata
(http://math-atlas.sourceforge.net/errata.html) you will see the
following snippet.

>===============================================
Gcc compiler bug hurts PowerPC Performance.

ATLAS 3.8 assumes gcc 4.2 (most recent gcc stable release at ATLAS
release). Gcc 4.2 is has a performance bug that normally cuts
performance in half. ATLAS works around this by throwing the flags:

   -fno-schedule-insns -fno-rerun-loop-opt

This makes the performance drop caused by using gcc-4.2 roughly 1-7%,
instead of 50%. If you are using a gcc version prior to gcc-4.2,
remove the these flags from all your flag macros in your
BLDdir/Make.inc after configure but before install to get back this
performance. If you are using a later gcc release, you may be able to
see if this bug has been fixed by scoping my original bug report .
>===============================================

I have no idea whether gcc-4.3.1 fixes the problem.  I have not tried
compiling gcc and gfortran since gcc-4.1.1/gfortran appeared on YDL.
>From memory this report seemed to mostly affect ppc970 and power5
processors.  However, it may be worth following up just to be sure of
the situation.  On a G4 machine it does not seem to be a problem.

-- 
Stephen Harker                           s.harker at adfa.edu.au
PEMS
UNSW at ADFA