This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: [Patch, fortran] PR28005 - gfortran: matmul produces wrong result

From: Janne Blomqvist <jblomqvi at cc dot hut dot fi>
To: Paul Thomas <paulthomas2 at wanadoo dot fr>
Cc: FX Coudert <fxcoudert at gmail dot com>, patch <gcc-patches at gcc dot gnu dot org>, "'fortran at gcc dot gnu dot org'" <fortran at gcc dot gnu dot org>
Date: Sat, 24 Jun 2006 23:39:11 +0300
Subject: Re: [Patch, fortran] PR28005 - gfortran: matmul produces wrong result
References: <44903FEE.2000705@wanadoo.fr> <44955298.6090708@gmail.com> <449B67FA.9090501@wanadoo.fr>

On Fri, Jun 23, 2006 at 06:03:06AM +0200, Paul Thomas wrote:
> FX Coudert wrote:
> 
> >2006-06-14  Paul Thomas  <pault@gcc.gnu.org>
> >
> >>
> >>    PR fortran/28005
> >>    * gfortran.dg/matmul_3.f90: New test.
> >
> >
> >OK.
> 
> This patch did not apply to 4.1 at all.   The  computational core has 
> diverged totally.  Since this PR wasa numerical error on good code, I 
> suggest that we should do something about it.  Would there be any 
> objection if I copied trunk's matmul into 4.1's?

The 4.1 compiler can IIRC still generate code that sets stride=0. The
check and correction for this was removed from trunk, so perhaps you
should take care of including these again if you want to backport to
4.1.

-- 
Janne Blomqvist

Attachment: pgp00000.pgp
Description: PGP signature

References:
- [Patch, fortran] PR28005 - gfortran: matmul produces wrong result
  - From: Paul Thomas
- Re: [Patch, fortran] PR28005 - gfortran: matmul produces wrong result
  - From: FX Coudert
- Re: [Patch, fortran] PR28005 - gfortran: matmul produces wrong result
  - From: Paul Thomas

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]