GitLab

The configure.ac of ELPA_2013.11 is cleaned up a bit
Also, ELPA_2013.11 is copied to ELPA_2014.06 in order to have
the base for the next ELPA release

By mistake the complex kernel with blocking 2 was not called
in parallel if OpenMP was used.

Due to an error in a preprocessor statement, the results for
real matrices were wrong if the kernels "avx-real-block6" or
"avx-real-block4" were chosen. No other kernels are affected.

The test programms always correctly stated that the results for
these kernels are wrong.

The Intel Fortran compiler accepts the flag "-fopenmp" for compilation
with OpenMP. However, the Intel MPI compiler wrapper does not.
With the Intel compiler, this leads to the fact, that if ELPA is compiled
with the "-fopenmp" flag a not thread-save version of the Intel MPI
library is used and the test (with make check) fails.

Intel promised to solve this in the future.

However, for now the problem is solved in the user friendly way that no
manipulation of the MPI compiler wrappers have to be done:

For detecting the OpenMP compiler flags, instead of the predefined
macro "AC_OPENMP" of autoconf a modified macro "AX_ELPA_OPENMP"
is used, which first checks "-openmp" and only then "-fopenmp".
Thus it is ensured that the Intel compiler (and mpi compiler wrapper)
does not get confused.

This is invisible for users calling "configure" during the installation
process.

If the ELPA library is compiled with OpenMP, the tests check
whether the MPI library provides the neccessary threading level.

There has been the error, that if the required threading level
was not available the test programs aborted, put no explicit
error code was set. This is changed now.

It is more user friendly if the configure steps aborts in the case
that the user did not define which ELPA kernel to use

A few kernels could not be used together with OpenMP:
an interface error prohibited the compilation.

Removed a forgotten C-preprocessor error flag.
Futhermore, describe in the documentation why at the moment a
build of Bluegene P/Q kernels and OpenMP is not allowed.

Automake does normally not understand the dependencies of Fortran 90
modules, thus a parallel build will almost certainly fail.

As in ELPA 2011.12.002, the ELPA_development_version_qr  now
includes scripts from the fdep project (see git.schokokeks.org/fdep)
which cirumvents this problem.

Thanks to Lorenz Huedepohl, now a parallel build works.

- updated INSTALL documentation
- improved description of kernels src/elpa2_kernels/README_elpa2_kernels.txt
- improved OpenMP builds:
   * usage of OpenMP can now be specified with
     "--enable-openmp" or "--with-openmp"
- improved checks for AVX build (identical to ELPA 2013.08.003):
   * if AVX kernels are specified AVX compilation is tested
   * the option "--with-avx-optimization" automatically updates
     CFLAGS and CXXFLAGS with optimization flags

This is the release of the ELPA_development_version_OpenMP
If OpenMP support is not used, this version has the same
functionality as ELPA_2013.08.

If OpenMP support is used, obviously, a hybrid version of ELPA will be
build.

Allthough this is a release, version ELPA_2013.11 is far from complete!
During the next week optimizations of the OpenMP part will be published,
however, the basic functionality is set by this commit