LAPACK Archives

[Lapack] Patches to test suite of BLAS for LAPACK-3.4.0

Dear Harald,

I echo Julien's thanks.

Just to say that you should beware of switches such as -Ofast in 
numerical work.  For example, with gfortran it specifically says
"Disregard strict standards compliance."

So, for instance, carefully placed brackets might be ignored thus 
perhaps damaging numerical stability.

Best wishes,

Sven Hammarling.

On 18/01/12 22:40, Harald Anlauf wrote:

some tests of the BLAS fail on Intel and AMD processors when
compiling with high optimization level (e.g. gfortran -Ofast).
The reason is that the determination of machine precision in
these tests only works without optimization.

I recommend to always use the Fortran 90+ intrinsic EPSILON()
where appropriate.

See the attached for details.


Lapack mailing list

<Prev in Thread] Current Thread [Next in Thread>

For additional information you may use the LAPACK/ScaLAPACK Forum.
Or one of the mailing lists, or