[Debichem-devel] GROMACS 4.6-beta1 release

[Mark Abraham]

On Tue, Dec 18, 2012 at 9:18 AM, Susi Lehtola <
jussilehtola at fedoraproject.org> wrote:

> On Fri, 30 Nov 2012 19:15:38 +0100
> Mark Abraham <mark.j.abraham at gmail.com> wrote:
> > Hello Jussi, Klaus, Nicholas and Christoph,
> >
> > I understand you guys manage the distribution of GROMACS packages for
> > the major Linux distributions.
> >
> > We've just made a beta release for GROMACS 4.6, and anticipate a few
> > more over the coming weeks and then a final release in the next two
> > months or so. We'd love for the packages to make their way out into
> > user space so we can get some feedback and uptake. We have made a
> > major shift to using CMake rather than autotools for our building and
> > packaging, which probably means the existing packaging schemes will
> > need some updating. Do let us know if we can help with details or
> > advice there.
> >
> > Source tarball:
> > ftp://ftp.gromacs.org/pub/gromacs/gromacs-4.6-beta1.tar.gz Manual
> > PDF: ftp://ftp.gromacs.org/pub/manual/gromacs-manual-4.6-beta1.pdf
> > Install guide:
> > http://www.gromacs.org/Documentation/Installation_Instructions
>
> Hi,
>
>
> first of all, please use gromacs-owner at fedoraproject.org as the contact
> address, since it will automatically go to all maintainers.
>

Noted, thanks.


>
> I squeezed in the time to have a look at this. I have been using CMake
> to build Gromacs in Fedora for some time now, so there are no problems
> on that end.
>
> However, GMX 4.6b1 fails to compile on Fedora 18 (GCC 4.7.2):
>
>
> gromacs-4.6-beta1/src/gmxlib/nonbonded/nb_kernel_avx_256_double/kernelutil_x86_avx_256_double.h:80:5:
> error: incompatible type for argument 1 of 'gmx_mm256_set_m128'
>

Yes, thanks. We know about that one: http://redmine.gromacs.org/issues/1074

On another note, it seems that the cmake script detects the CPU on the
> compile system and produces an optimized version for the specific CPU.
> Naturally this is somewhat problematic for distributions, since e.g. an
> x86_64 binary should run on all x86_64 systems. Is this the case or not?
>

The default build produces an executable that targets the best instruction
set supported by the build host, and the executables will not run if that
set is not supported. From the point of view of packaging, that's ugly.
Either there needs to be multiple packages, or you will need to choose an
instruction set that is conservative enough to always work.
-DGMX_CPU_ACCELERATION=None will always work, but will be horribly
slow. -DGMX_CPU_ACCELERATION=SSE2 probably runs about as fast as the 4.x
series, but is probably faster through our better use of force-only
kernels. -DGMX_CPU_ACCELERATION=SSE4.1 is only slightly faster than SSE2.
And then there is AVX, which is faster still, but stil having teething
issues.

I'd suggest having an SSE2 and a None, but it really depends on what the
downstream people want.

Tomorrow, I'll make some notes in our installation instructions about these
issues.

Mark

--
> Susi Lehtola
> Fedora Project Contributor
> jussilehtola at fedoraproject.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/debichem-devel/attachments/20121220/20e94d09/attachment.html>