Re: Changing octave to exploit multi-core hardware

[Peter L.]

ATLAS can be compiled to be multithreaded. Just turn it on, and this
should give you multithreaded BLAS and LAPACK with very little effort.



tir, 25 03 2008 kl. 15:19 +0200, skrev Shai Ayal:
> On Tue, Mar 25, 2008 at 3:02 PM, Jaroslav Hajek <address@hidden> wrote:
> >
> >  A number of vendor-tuned BLAS libraries are multithreaded, usually
> >  along with key BLAS subroutines (triangular & QR factorizations etc).
> >  LAPACK itself is not, AFAIK.
> >  There is no problem using these MTed BLAS libs (ACML, Intel MKL, Goto
> >  BLAS) with Octave to exploit multiple cores when working with large
> >  matrices.
> 
> While there might be no problem using these libs, there is a problem
> in redistributing them as none of them are GPL compatible, and some of
> them have restrictions on some uses (e.g. Goto is non-free for
> commercial use)
> 
> Shai