[gentoo-dev] proposing new virtuals: blas and lapac + new use flag: ifc

[gentoo-dev] proposing new virtuals: blas and lapac + new use flag: ifc

[George Shapovalov]

Hi gang.

Courtezy of Derek Dolney (not a dev) we have an improvement proposed towards 
some of the sci libs. But first a bit of explanation.
(#30453 has all the details)

BLAS is a Basic Linear Algebra System - is a famous and really basic library. 
Not surprisingly, there are a couple of implementations - first a reference 
one, completely in fortran and called just that - blas. Another often used 
one is atlas (and there are more, for example gsl comes with cblas library as 
part of it).

These libraries can be substituted for one another in majority of cases. As 
such it would be nice to have a virtual/blas and correspondingly blas-config 
utility.

Lacpack is another library (Linear Algebra PACKage) that provides more usefull 
features, It depends on blas (either blas or atlas will do), so is one of the 
would be users of virtual/blas and also has a reference and optimized (atlas 
based) implementation. It is also usefull to be able to have both installed 
together (to do chacks of validity of optimized runs for example). As such 
this would require another virtual: virtual/lapack.

After discussing this with Derek we converged on the following naming scheme 
(does not require approval really, but may be interesting for blas/lapack 
users):
blas-reference and blas-atlas, plus lapack-reference and lapack-atlas
which should make sense for all blas/lapack users (now is your time to object 
if it doesn't ;)).

Now on to the last part: ifc use flag:
So, to complicate things a bit more ;), these libraries (their fortran parts) 
can be compiled with either g77 of ifc (Intel fortran compiler - unrestricted 
download btw, as opposed to icc). However, as these are quite different apps 
name and ionvocation wise, yet another virtual is inappropriate here (and 
wouldn't belong to sci herd legacy anyway), but a use flag is just right.

So, hereby I am proposing above described extensions to blas-related lisb that 
require two new virtuals: blas and lapack and new use flag ifc.

George



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] proposing new virtuals: blas and lapac + new use flag: ifc

[splite-gentoo]

On Wed, Oct 29, 2003 at 11:17:57AM -0800, George Shapovalov wrote:
>[...]
> can be compiled with either g77 of ifc (Intel fortran compiler - unrestricted 
> download btw, as opposed to icc). However, as these are quite different apps 
>[...]

The Intel Fortran Compiler for Linux (ifc) is definitely not an unrestricted
download.  You may download and use the compiler under a no-cost
"Non-Commercial Unsupported" license* that essentially covers running
it on your home computer for your own amusement.  (I'm not sure how many
people use BLAS or LAPACK strictly for their own amusement. :)  Paid academics
(faculty, grad students) aren't covered, even if their research is not for
profit.

* http://developer.intel.com/software/products/compilers/flin/flin_noncom_lic.htm

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] proposing new virtuals: blas and lapac + new use flag: ifc

[George Shapovalov]

On Thursday 30 October 2003 13:08, splite-gentoo@sigint.cs.purdue.edu wrote:
> On Wed, Oct 29, 2003 at 11:17:57AM -0800, George Shapovalov wrote:
> The Intel Fortran Compiler for Linux (ifc) is definitely not an
> unrestricted download.  You may download and use the compiler under a
> no-cost
> "Non-Commercial Unsupported" license* that essentially covers running
> it on your home computer for your own amusement.  (I'm not sure how many
> people use BLAS or LAPACK strictly for their own amusement. :)  Paid
> academics (faculty, grad students) aren't covered, even if their research
> is not for profit.

Hm, thanks for clarification! (having not dealt with ifc itself, just testing 
blass libs and seeing how it was automatically pulled off the mirrors I 
assumed it is less restricted.)
So, shouldn't the ifc ebuild be fetch-restricted then (just like the icc one)? 
Doesn't Intel want to enforce a usual click-through thing (to which this page 
actually points)?

George



--
gentoo-dev@gentoo.org mailing list