Boost logo

Glas :

Re: [glas] dense and sparse vectors

From: C J Kenneth Tan -- OptimaNumerics (cjtan_at_[hidden])
Date: 2005-08-05 08:56:17

Toon, Karl and others,

On 2005-08-05 15:03 +0200 Toon Knapen (toon.knapen_at_[hidden]) wrote:

> > Which algorithms are going to use collections? If we answer this
> > question first, we can also decide which iterators are needed to
> > implement these algorithms? (We may conclude that iterators are not good
> > at all).
> >
> > One example given by Toon is printing a vector.
> Indeed. As Andrew Lumsdaine remarked some time ago: we should look at
> what's necessary to implement the algorithms.
> Therefore I will next implement a sparse and dense axpy and even try to
> benchmark this against other libraries. This might reveal a lot.

The benchmarks of dense daxpy and dgemm, running against system
vendors' implementations, will be interesting to see.

More interesting would be higher level code, such as dposv, dppsv,
dgesv, dgeev, dgesvd, dsyev, dgels, or their equivalents. While daxpy
and dgemm code are relatively straight forward, tuning these higher
level code is often non-trivial, especially if the code is to be
platform independent.

Kenneth Tan
News: OptimaNumerics Enhances Performance on SGI Altix Supercomputers
C J Kenneth Tan, PhD
OptimaNumerics Ltd Telephone: +44 798 941 7838
E-mail: cjtan_at_[hidden] Telephone: +44 871 504 3328
Web: Facsimile: +44 289 066 3015