[SciPy-user] numarray interface and performance issues (for dot product and transpose)

eric eric at scipy.org
Thu Feb 28 15:25:08 CST 2002


It should be possibly to just reimplement the NumPy functions like dot and
matrixmultiply (name it matmul like F90 or mat_mul) within SciPy to use ATLAS.
Then, using the following:

    from scipy import *
    ...
    c = dot(a,b)

should go fast.  The NumPy guys *might* go for requireing ATLAS, but I find that
highly unlikely because it is such a bear to build on some platforms and takes
so long on all platforms.  One of the beauties of NumPy is that it is self
contained.

SciPy has already bitten the bullet as far as considering ATLAS as part of the
core requirements, so it comes as no extra effort.  We should just add the ATLAS
code patch to SciPy's linalg module and then import those functions into the
main namespace.

eric

----- Original Message -----
From: "R.M.Everson" <R.M.Everson at exeter.ac.uk>
To: <scipy-user at scipy.org>
Sent: Thursday, February 28, 2002 4:46 PM
Subject: Re: [SciPy-user] numarray interface and performance issues (for dot
product and transpose)


>
> Alex,
>
> Seems like you have generated some discussion at least.  Perhaps we
> should download the new numarray and see what's involved in ATLASifying
> it.
>
> Best,
> Richard.
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.net
> http://www.scipy.net/mailman/listinfo/scipy-user
>





More information about the SciPy-user mailing list