Atlas compilation

[Brooks Davis]

On Wed, Oct 13, 2004 at 09:54:09AM -0700, Kris Kennaway wrote:
> On Wed, Oct 13, 2004 at 07:38:44AM -0700, Brooks Davis wrote:
> > On Tue, Oct 12, 2004 at 08:48:56AM -0700, Kris Kennaway wrote:
> > > On Mon, Oct 11, 2004 at 03:05:20PM +0200, Toyotoshy wrote:
> > > > Hallo evrybody,
> > > > I'm just a newbie and I'm trying to compile /usr/ports/math/atlas
> > > > Compilation started 6 hours ago and:
> > > > 
> > > > du /usr/ports/math/atlas
> > > > ....
> > > > 360160  /usr/ports/math/atlas
> > > > 
> > > > isn't it a bit too much time and space?
> > > 
> > > No, it's an enormous port.  Depending on how slow your machine is, it
> > > may take days to build, or it could be as little as 24 hours.
> > 
> > Yes, not only does it do a lot of things, but since it is
> > Automatically Tuned Linear Algebra Software, it actually compiles
> > several versions of each function and runs microbenchmarks on them to
> > determine which one to use.  Ideally, you should build it when the
> > system is nearly idle to avoid perturbing the results.
> 
> Hmm, that probably means the package is useless, right?

Not entirely given the build time, but certainly if you want to use it
seriously, the package isn't all that useful.

-- Brooks

-- 
Any statement of the form "X is the one, true Y" is FALSE.
PGP fingerprint 655D 519C 26A7 82E7 2529  9BF0 5D8E 8BE9 F238 1AD4
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-ports/attachments/20041013/35f2f33b/attachment.bin