Bug#569436: [Pkg-scicomp-devel] Bug#569436: Bug#569436: lapack: zgesvd seems to give incorrect results

[Kumar Appaiah]

On Mon, Feb 15, 2010 at 10:39:52AM -0600, Kumar Appaiah wrote:
> On Sat, Feb 13, 2010 at 09:59:24AM -0600, Kumar Appaiah wrote:
> > > * Explicit dependency on gcc 4.3 removed (Closes: #533807)
> > > but the changelog does not say what this explicit dep has been added...
> > 
> > I would also suspect that this is a compiler related issue; for your
> > modifications are only in the control file and changelog. I'll try to
> > probe this further, and let you know.
> 
> Upon further probing, I can confirm that if I switch the -O3's in
> debian/rules to -O2, then things start working fine. I'll try to probe
> further and see how to put the blame elsewhere (on the compiler, that
> is).

OK, after wasting several hours on this, I couldn't really get to the
root of the problem. The trouble is that zgesvd causes several lapack
calls, each of which calls several lapack functions; it is very
difficult to isolate which Blas function actually causes the error.

Since SVD is something I consider very important, I would request you
to please re-upload Blas with an optimization level of -O2, so that
things remain sane with all the reverse dependencies.

Thanks, and sorry.

Kumar
-- 
But what can you do with it?  -- ubiquitous cry from Linux-user partner.
(Submitted by Andy Pearce, ajp at hpopd.pwd.hp.com)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/debian-science-maintainers/attachments/20100215/d0997373/attachment-0001.pgp>