Re: [Help-gsl] Question about solving linear equation with GSL

[Brian Gough]

At Tue, 6 Jan 2009 16:25:17 -0500,
Sheng Yu wrote:
> It takes 0.25 seconds for the code to do Cholesky Decomposition and
> solve. However, when I do it with Matlab with A\b, it only
> takes 0.045 seonds (It even does not take advantage of the fact that A
> is symmetric and positive definite). I'm wondering why
> Matlab is much faster. Did I do something wrong?

Try linking with Atlas for the cblas (described in the manual), it may
give you some speedup since the routine is all calls to BLAS ddot.  

We don't use a block algorithm, so we won't be as fast as LAPACK
(which is what Matlab uses).

-- 
Brian Gough
(GSL Maintainer)

Support freedom by joining the FSF!
http://www.fsf.org/associate/support_freedom/join_fsf?referrer=37