[Pw_forum] bug in atlas

[Axel Kohlmeyer]

On Tue, 23 Jan 2007, Konstantin Kudin wrote:

KK>  Dear Prashant,
KK> 
KK>  The question you are asking should be straightforward to figure out on
KK> your own. Just grep through the source code (it is GPL and is
KK> available), find all the instances of complex GEMM, and find those that
KK> utilize the special case.
KK> 
KK>  Then you can report your findings back to the list.

good idea! and very useful, too. 

but sadly, that is only (the easy) half of the job. you also 
have to check the LAPACK subroutines that QE uses for the use 
of that problematic call, which is a bit messy since there
are quite a few cross-calls from within lapack, so best
is to check all of it.

cheers,
   axel.

KK> 
KK>  Regards,
KK>  Kostya
KK> 
KK> 
KK> --- Prashant Shevgaonkar <shevgaonkar at gmail.com> wrote:
KK> 
KK> > Atlas has recently (20Nov06) reported an error in computation of
KK> > complex
KK> > GEMM (C = A*B +C) for the special case B = Atransp.
KK> > (
KK> >
KK> http://sourceforge.net/mailarchive/forum.php?thread_id=31031975&forum_id=434)
KK> > How serious is this error and would this affect pwscf calculations?
KK> > 
KK> > ------------------------------------------------------------------
KK> > Prashant Shevgaonkar,
KK> > Chemical Engineering Department,
KK> > Vishwakarma Institute of Technology,
KK> > Pune, India 411 037
KK> > Email: shevgaonkar at gmail.com
KK> > 
KK> 
KK> 
KK> 
KK>  
KK> ____________________________________________________________________________________
KK> Yahoo! Music Unlimited
KK> Access over 1 million songs.
KK> http://music.yahoo.com/unlimited
KK> _______________________________________________
KK> Pw_forum mailing list
KK> Pw_forum at pwscf.org
KK> http://www.democritos.it/mailman/listinfo/pw_forum
KK> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.