64-bit cleanlinest of gfortran

[Steve Kargl]

On Tue, Dec 14, 2004 at 07:47:46PM +0100, Thomas Koenig wrote:
> Steve,
> 
> you already found PR 18977 where I hit a segfault with -O1 and -O3
> while running the Lapack testsuite on Itanium.
> 
> The Lapack testsuite did run at -O0, but the results were quite
> disappointing.  I have added them to PR 5900 (which seems to be
> a collecting point of Lapack test results to me).
> 

Scott Robert Ladd suggested that gcc is generating bad SSE/SSE2
code, and SSE/SSE2 is the default on amd64.  I've tried setting
-mfpmath=387, but I still get several BUS ERROR and failed test. 

-- 
Steve