Gentoo Archives: gentoo-science

From: "M. Edward (Ed) Borasky" <znmeb@×××××××.net>
To: gentoo-science@l.g.o
Subject: Re: [gentoo-science] [Fwd: BLAS and CBLAS in Gentoo]
Date: Tue, 10 Jul 2007 14:19:11
In Reply to: Re: [gentoo-science] [Fwd: BLAS and CBLAS in Gentoo] by Markus Dittrich
1 Markus Dittrich wrote:
2 > On Tue, 10 Jul 2007, Adam Pityszek wrote:
3 >> Hi All,
4 >
5 >> I am forwarding my email to Sebastien regarding the merging of blas- and
6 >> cblas-reference into one package. Sebastien accepted this idea, but we
7 >> would like to hear your opinions as well, since you are the real users
8 >> of scientific libraries in Gentoo.
9 >
10 >> BR,
11 >> /Adam
12 >
13 >
14 > This sounds like a fine plan to me, particularly since most
15 > blas implementations in portage do provide both blas and cblas
16 > at the same time. In addition, this would render the eselect framework
17 > less complicated which I find desirable.
18 >
19 > cheers,
20 > Markus
21 >
22 >
23 > -- Markus Dittrich (markusle)
24 > Gentoo Linux Developer
25 > Scientific applications
27 I think the general scenario for users of BLAS and LAPACK is that they
28 will be happy with this. However, some people, especially in the
29 high-performance area, will want to be able to benchmark various
30 versions (ATLAS, GOTO or chip-specific Intel/AMD) and either static-link
31 to the winner or dynamic-link to the winner on an individual calling
32 application basis.
34 I haven't gotten around to doing that yet, but it's on my list for my
35 AMD64 dual-core system. I'm not sure the "eselect" facility provides for
36 this. And a number of packages carry their own BLAS or LAPACK. R is one
37 of them; in fact
39 1. R allows you to build a shared library of the BLAS and LAPACK that it
40 carries, and
41 2. The R team strongly suggests using the LAPACK that R carries rather
42 than an external one, and indeed, the "make check-all" tests fail on
43 some architectures with some external versions of LAPACK. On my Athlon
44 T-Bird, this happened with the Gentoo lapack-atlas and the R team
45 slapped me silly when I complained about it on their mailing list. ;)
47 So as long as I can do all of these things:
49 1. Choose between static and dynamic linking to BLAS/LAPACK with a USE
50 flag on any application that allows such a choice,
51 2. Choose between internal BLAS/LAPACK with a USE flag on any
52 application that allows such a choice, and
53 3. Choose between the various versions for the highest performance on an
54 individual application basis
56 I'm happy. I think most of the ebuilds are already set up this way, but
57 again, I haven't had a chance to test with really big problems on my
58 really big machine yet.
59 --
60 gentoo-science@g.o mailing list