2014-05-21 14:08 GMT+02:00 Lars Buitinck :
> 2014-05-21 13:59 GMT+02:00 Sergio Pascual :
> > This is the patch we use in fedora to compile scikit-learn 0.14.1 with
> > system cblas
> >
> >
> http://pkgs.fedoraproject.org/cgit/python-scikit-learn.git/tree/sklearn-unbundle-cblas.patch
>
> But the bu
Sturla Molden
wrote:
> It is only the Fortran BLAS that uses f2c ABI, not cblas. SciPy now has a
> fix for Accelerate: SciPy has Fortran BLAS and LAPACK wrappers on top
> Accerate's C interface.
Well, if BLAS is built with g77 on other platforms (e.g. Windows) it might
be affected as well.
Bu
Sergio Pascual
wrote:
> This is the patch we use in fedora to compile scikit-learn 0.14.1 with
> system cblas
>
> href="http://pkgs.fedoraproject.org/cgit/python-scikit-learn.git/tree/sklearn-unbundle-cblas.patch";>http://pkgs.fedoraproject.org/cgit/python-scikit-learn.git/tree/sklearn-unbundle-
Skipper Seabold wrote:
> Pauli pointed me here for the troublesome ABIs with Apple's Accelerate [2].
It is only the Fortran BLAS that uses f2c ABI, not cblas. SciPy now has a
fix for Accelerate: SciPy has Fortran BLAS and LAPACK wrappers on top
Accerate's C interface.
There is no longer an ABI
On Wed, May 21, 2014 at 8:00 AM, Lars Buitinck wrote:
> 2014-05-21 13:47 GMT+02:00 Olivier Grisel :
>> This is a great trick. We might want to get rid of our own partial
>> copy of CBLAS at some point.
>
> I remember Radim (gensim maint) describing some trouble with BLAS ABIs
> on a mailing list s
2014-05-21 13:59 GMT+02:00 Sergio Pascual :
> This is the patch we use in fedora to compile scikit-learn 0.14.1 with
> system cblas
>
> http://pkgs.fedoraproject.org/cgit/python-scikit-learn.git/tree/sklearn-unbundle-cblas.patch
But the build system should skip our copy if CBLAS headers and libs
a
2014-05-21 13:47 GMT+02:00 Olivier Grisel :
> This is a great trick. We might want to get rid of our own partial
> copy of CBLAS at some point.
I remember Radim (gensim maint) describing some trouble with BLAS ABIs
on a mailing list some time ago, but I can't find the mail right now.
Is this the s
This is the patch we use in fedora to compile scikit-learn 0.14.1 with
system cblas
http://pkgs.fedoraproject.org/cgit/python-scikit-learn.git/tree/sklearn-unbundle-cblas.patch
2014-05-21 13:47 GMT+02:00 Olivier Grisel :
> This is a great trick. We might want to get rid of our own partial
> cop
This is a great trick. We might want to get rid of our own partial
copy of CBLAS at some point.
--
Olivier
--
"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests acros
Lars Buitinck wrote:
> I think this question comes about because Manoj is trying to optimize
> some Cython code. But the question "do you have the exact same
> library" is valid.
Well, this approach allows us to call BLAS and LAPACK with the overhead of
a function pointer, so it's rather cheap :
The error message shows the system is looking in the system library
directories for cblas. When you changed the join statement, and probably
when you started working in a different directory, the build script no
longer knows how to find the cblas library. So you need to make the join
statement co
I wanted to reproduce the behaviour of ddot outside the scikit-learn
directory, so that I can actually see what is happening.
This is what I did.
1. Copied the cblas files from sklearn/src to a directory called CyPractise
2. I looked at the setup.py file over here (
https://github.com/scikit-le
2014-04-27 18:53 GMT+02:00 Sturla Molden :
> Unlike the NumPy _dotblas module, SciPy uses an f2py wrapper that actually
> exports a function poiinter.
>
> Using this scheme to code a fake cblas layer is not difficult either.
I think this question comes about because Manoj is trying to optimize
som
Manoj Kumar
wrote:
> I want to use cblas libraries outside the sklearn directory, but I have
> trouble in understanding how to set it up. It keeps returning the error,
> /usr/bin/ld: cannot find -lcblas .
Do you actually have the libcblas used to build NumPy?
If you are happy with Fortran BLAS
Hello,
I want to use cblas libraries outside the sklearn directory, but I have
trouble in understanding how to set it up. It keeps returning the error,
/usr/bin/ld: cannot find -lcblas . I took a cue from the sklearn setup.py
files and did in this in the setup.py file. Not that the cython file is
15 matches
Mail list logo
