Hi Jean-Pierre and all,
> Grat!
> At some point, I pushed patches on linbox-devel (or something like that), but
> never got any reply.
> IIRC they were just tiny patches to ease ocmpilation on unusual archs.
> I did not check recently, but one year ago or so they did not make it
> upstream.
> It
Hi Clément and all,
On Friday, August 29, 2014 10:53:18 AM UTC+2, Clement Pernet wrote:
>
> Hi,
>
> Let me clarify a few things:
> - LinBox's approach to link against numerical BLAS has never changed, and
> probably will not in the
> near future.
> - LAPACK is not BLAS: BLAS provides optimiz
On Sun, Aug 31, 2014 at 3:16 PM, Volker Braun wrote:
> You misunderstood, we are not going to distribute Sage linked to MKL. This
> is obviously not legal to distribute. But it should be possible for you to
> compile Sage with alternate BLAS implementations on your own machine if you
> chose to.
You misunderstood, we are not going to distribute Sage linked to MKL. This
is obviously not legal to distribute. But it should be possible for you to
compile Sage with alternate BLAS implementations on your own machine if you
chose to.
On Sunday, August 31, 2014 10:53:49 PM UTC+1, Fernando P
On Fri, Aug 29, 2014 at 3:24 AM, Volker Braun wrote:
> I do have a MKL license (Linux + OSX) for the Sage project
Who is going to assume the indemnification liability on behalf of Sage if
MKL-linked binaries are distributed?
And even if someone does, it will require some form of license mo
How do you deal with OpenBLAS not supporting fork? They use OpenMP and the
gcc implementation does not allow fork(), at least for now. If you use
OpenBLAS before forking, then your program will deadlock in OpenBLAS after
the fork.
On Friday, August 29, 2014 10:55:32 AM UTC+1, François wrote:
William A Stein wrote:
> What's the situation with speed regression testing in Sage? A couple
> of years ago I think people wrote a regression testing framework
> (maybe David Roe or Robert Bradshaw?)
FWIW I started a little bit of work aiming to make the existing
benchmarking code more useful a
On Friday, August 29, 2014 11:28:47 AM UTC+1, François wrote:
>
> OK. How do we benchmark?
This is the real question. We should have a set of benchmark tests that are
longer than ordinary doctests (and not part of the testsuite) that mostly
rely on the underlying blas.
--
You received this m
OK. How do we benchmark? Actually I haven't ever linked the whole thing
against a proprietary lib. In theory it all works but I never tested that bit
in
the field.
On 29/08/2014, at 22:24, Volker Braun wrote:
> I do have a MKL license (Linux + OSX) for the Sage project
>
>
>
> On Friday
I do have a MKL license (Linux + OSX) for the Sage project
On Friday, August 29, 2014 10:55:32 AM UTC+1, François wrote:
>
> I currently drive my sage install with openblas, it would be interesting
> if someone
> has a MKL license.
>
--
You received this message because you are subscrib
Volker and me had some kind of plan to get sage to to compile with any
blas/lapack
based on providing .pc file for blas/cblas/lapack. It is already 100% done in
sage-on-gentoo
and I have the patches to convert the sage library.
I currently drive my sage install with openblas, it would be interest
On 2014-08-29 10:53, Clement Pernet wrote:
http://www.openblas.net/
The compile & configuration times are much shorter which would impact sage
compile time.
+100 to getting rid of ATLAS in Sage!
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
Hi,
Let me clarify a few things:
- LinBox's approach to link against numerical BLAS has never changed, and
probably will not in the
near future.
- LAPACK is not BLAS: BLAS provides optimized numerical matrix multiplication
kernels (that LinBox
uses) and other basic routines, whereas LAPACK provi
On Fri, Aug 29, 2014 at 10:24 AM, Jean-Pierre Flori wrote:
> I don't know for sure, but I think linbox does not link to LAPACK
> (anymore?).
> Maybe that make a small difference already.
That's precisely what I'm worried about -- not linking lapack would
perhaps change linbox from being world cl
I don't know for sure, but I think linbox does not link to LAPACK
(anymore?).
Maybe that make a small difference already.
I opened a ticket about that some monthes ago.
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this grou
15 matches
Mail list logo