The princomp routine on octave forge uses
C = cov(X)
[U,D,pc] = svd(C,1)
Can you tell us which of these is running slower on large arrays?
How does the performance of other routines such as
lu, A*B and qr compare?
Does biG5 mean dual processor? Does veclib automatically
parallelize blas? Are both processors active?
Paul Kienzle
address@hidden
On Jun 21, 2004, at 10:22 AM, Samir Sharshar wrote:
Hi,
Please let me report my experience about Octave-2.1.57 compiling and
installing on a Mac OS 10.3.4 (Panther) with the last development
tools from Apple that provides the vecLib libraries and headers (a
BLAS, LAPACK like, as I mean).
I've compiling and installing Octave-2.1.57 from sources with prior
installation of g77-3.5 and fftw3.0.1-fma.
Octave runs perfectly on my machine; a biG5 at 2 gHz with 4 Go RAM.
However, there is a problem that I can't resolve by myself. Principal
component analysis runs really slowly (three times less than on a
equivalent octave installation on a Debian runnng on a PIV 3.2 GHz).
Things are worst when octave was compiling linking ATLAS libraries
instead of vecLib ones.
Have you any idea ?
Thanks in advance,
Best wishes,
Samir Sharshar MD
Inserm, Lille, France
-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.
Octave's home on the web: http://www.octave.org
How to fund new projects: http://www.octave.org/funding.html
Subscription information: http://www.octave.org/archive.html
-------------------------------------------------------------
-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.
Octave's home on the web: http://www.octave.org
How to fund new projects: http://www.octave.org/funding.html
Subscription information: http://www.octave.org/archive.html
-------------------------------------------------------------