Mailing List Archive

intel packages support
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

I am in the process of testing updates of dev-lang/icc dev-lang/icc and
sci-libs/mkl on the tree. Right now, I just put an update of these
packages together with sci-libs/ipp in the gentooscience overlay.

The Linux versions of a bunch of Intel products come with non-commercial
licenses [1], but this license is quite restrictive [2]. So before I
carry on more testing and push them to the tree, I would like feed-back
on the following points.

- - I've seen various mentions or bugs request to remove Intel products
from the tree. Are we willing to keep these packages?
Keeping up with updates and bugs is not trivial mainly because upstream
always changes packaging style and are large packages (sometimes 300Mb)

- - do you see a need, or significant performance increases with the
compilers? Up to gcc/gfortran-4.2, ifc was the only complete FORTRAN-95
compiler in portage, so there was a definite need among the scientific
community.

- - in the past, they were fetch-restricted. I don't really see why in the
documents I've read mirror-restricted should not be enough.

Obviously anyone willing to help on the process is welcome.

Regards,

- --
Sébastien

[1] http://www.intel.com/cd/software/products/asmo-na/eng/340679.htm
[2] http://www.intel.com/cd/software/products/asmo-na/eng/219692.htm


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFG1uC41ycZbhPLE2ARAlRnAJ94enzEM6mcf4o22WLW9q3JO/T5KwCdHKs4
vTKxHAGfXQQACXdkAWLhUK4=
=tFfH
-----END PGP SIGNATURE-----
--
gentoo-dev@gentoo.org mailing list
Re: intel packages support [ In reply to ]
On Thursday 30 August 2007, Sébastien Fabbro wrote:
> - I've seen various mentions or bugs request to remove Intel products
> from the tree. Are we willing to keep these packages?
> Keeping up with updates and bugs is not trivial mainly because upstream
> always changes packaging style and are large packages (sometimes 300Mb)
>
> - do you see a need, or significant performance increases with the
> compilers? Up to gcc/gfortran-4.2, ifc was the only complete FORTRAN-95
> compiler in portage, so there was a definite need among the scientific
> community.
>
> - in the past, they were fetch-restricted. I don't really see why in the
> documents I've read mirror-restricted should not be enough.

pretty much every question here is left up to the maintainer

if there is a dev out there who wants it in the tree and maintains it, then do
it
-mike
Re: intel packages support [ In reply to ]
> - - do you see a need, or significant performance increases with the
> compilers? Up to gcc/gfortran-4.2, ifc was the only complete FORTRAN-95
> compiler in portage, so there was a definite need among the scientific
> community.

Performance is probably the big thing. A few months ago I tested some
simple looping code with both ifc and gfortran. The ifc compiled
executable was significantly faster (don't recall how much right now).
I can only imagine the performance gains would be even better with
more complex code. That said, my stuff isn't performance critical and
the license situation is obviously much nicer with gfortran, so I just
use that anyway.

Also, I agree with Mike...if somebody wants to maintain it, why remove it?

-Steve
--
gentoo-dev@gentoo.org mailing list