[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] New release?
- From: Gael Guennebaud <gael.guennebaud@xxxxxxxxx>
- Date: Fri, 30 Sep 2011 08:14:03 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type:content-transfer-encoding; bh=ogRi1jeaF6XssGCGYjUUS0su7qztAzur8I+Gt4LJdh0=; b=v1EyXEyTu0Y3xSlkKXTKiMaJSJpuPa8vfTA8/nIFgPCIdMZKLwjUuflGhsIScfJnq5 NuXLHxsvL1wg2AONakWneeUlp195kq1Q3Cbb8iMcKt9MGl5PKrJoPMaMiQjE+p0ycAQ0 5PYQEJI91/YtFj3aBh2kD6H2Q40oYkLN65UEc=
btw the reason for that change is that such incomplete BLAS and Lapack
are still enough for SuperLU, UmfPack, Cholmod and the likes....
Gael
On Fri, Sep 30, 2011 at 8:12 AM, Gael Guennebaud
<gael.guennebaud@xxxxxxxxx> wrote:
> Hi,
>
> the 3.0 branch did not change with that respect ("hg up 3.0" to switch
> to it), however I changed the devel branch to enable the compilation
> of incomplete BLAS and Lapack libs without the functions still
> implemented in Fortran. Tried on a Linux+gcc and windows+MSVC with and
> without a fortran compiler.
>
>
> Gael.
>
> On Fri, Sep 30, 2011 at 7:03 AM, Pavel Holoborodko
> <pavel@xxxxxxxxxxxxxxx> wrote:
>> Hi,
>> I've tried to compile the most recent Eigen with MSVC 2010 SP1.
>> Cmake chokes up with:
>> Check for working Fortran compiler using: Visual Studio 10 -- broken
>> It gives error messages like "The Fortran project file (.vfproj) is corrupt"
>> just before stop working.
>> I have Intel Fortran Installed, test project compiles nicely with it.
>> Is it me or something has changed since 3.0.2 - which had been
>> compiled/tested on the same PC without any problems.
>> Does anybody else experienced this problem?
>> Thanks for all the hard work.
>> On Thu, Sep 29, 2011 at 12:50 AM, Gael Guennebaud
>> <gael.guennebaud@xxxxxxxxx> wrote:
>>>
>>> thanks for the report.
>>>
>>> already fixed.
>>>
>>> gael
>>>
>>> On Wed, Sep 28, 2011 at 4:33 PM, Rhys Ulerich <rhys.ulerich@xxxxxxxxx>
>>> wrote:
>>> >> Everybody, please run the tests as described on
>>> >> http://eigen.tuxfamily.org/index.php?title=Tests
>>> >>
>>> >> Do also run the Eigen2 tests by setting the cmake variable
>>> >> EIGEN_TEST_EIGEN2 to on as this went wrong with the previous release..
>>> >
>>> > With EIGEN_TEST_EIGEN2=1 the 3.0 branch (4259:01d81e17e78a) fails
>>> > 'make check' with a compilation error for both Intel 11.1 and Intel
>>> > 10.1. The errors start like the following for both compilers...
>>> >
>>> > [ 59%] Building CXX object
>>> >
>>> > test/eigen2/CMakeFiles/eigen2_prec_inverse_4x4_3.dir/eigen2_prec_inverse_4x4.cpp.o
>>> >
>>> >
>>> > /net/setun/workspace/rhys/tmp/eigen/Eigen/src/Core/arch/SSE/PacketMath.h(165):
>>> > error: namespace "Eigen::internal" has no member "copy_bool"
>>> > { eigen_assert(false && "packet integer division are not supported by
>>> > SSE");
>>> > ^
>>> >
>>> >
>>> > /net/setun/workspace/rhys/tmp/eigen/Eigen/src/Core/products/Parallelizer.h(37):
>>> > error: namespace "Eigen::internal" has no member "copy_bool"
>>> > eigen_internal_assert(v!=0);
>>> > ^
>>> >
>>> > Would y'all like me to roll up the details into a formal bug?
>>> >
>>> > - Rhys
>>> >
>>> >
>>> >
>>>
>>>
>>
>>
>