From: boost (boost_at_[hidden])
Date: 2002-06-28 15:22:34
On Friday 28 June 2002 11:37, Martin Weiser wrote:
> Even though it's fairly easy, a specialized version of prod() calling
> xgemm under the hood would be even easier to use. Insofar the interface
> is influenced by the ability to specialize core operations, which is why
> I started this thread during review in the first place.
Sure, I'd love to remove my #defines just for replacing prod with xgemm.
In addition this performance loss forced me to postpone my automatic
differentiation project, I announced in April here, as I can't call a
corresponding xgemm, and ublas is far too slow. Therefore it would
be really nice to have an ublas implementation with ATLAS performance, but
that's surely beyond an initial release.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk