Building DAMASK on cluster from Source against Centrally Installed PETSc and HDF5 #11
-
We're encountering issues while attempting to build DAMASK against our centrally installed versions of PETSC and HDF5. Our aim is to utilise these centrally installed modules as they are built against our MPI setup. However, during the build we are encountering errors such as |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 2 replies
-
Can you confirm that the Fortran bindings have been included when building your HDF5 library? |
Beta Was this translation helpful? Give feedback.
-
DAMASK relies on the PETSc build system and PETSc needs to be aware of HDF5 incl. Fortran bindings (see https://damask.mpie.de/support/FAQ/installation#why-does-compilation-of-grid-or-mesh-solver-fail-due-to-missing-files).
The Fortran code of HDF5 is in separate |
Beta Was this translation helpful? Give feedback.
DAMASK relies on the PETSc build system and PETSc needs to be aware of HDF5 incl. Fortran bindings (see https://damask.mpie.de/support/FAQ/installation#why-does-compilation-of-grid-or-mesh-solver-fail-due-to-missing-files).
If your PETSc is not aware of HDF5 (and FFTW) but you can ensure compatibility, installation requires manual tweaking of
CMakeLists.txt
(the version in the root directory, notsrc/CMakeLists.txt
). This file contains some inactive code marked as "experimental" that uses the FindPkgConfig feature of CMake to search for those packages.Manual modifi…