As for the issues I had with RBF mapping. I have installed PETSc v3.12.5. I am adding the install log, just in case.
But, I am still getting the same error when I do
Installation looks correct.
When you run
cmake on preCICE, does it find the correct PETSc version? You can check this in the output you get.
Could you also please confirm that without solving these issues if I try to
make install, the installation will fail?
No, these two things are completely independent. Installation of preCICE only means that the library and headers are copied somewhere else. Installation is also optional.
Because if I try to
make install I get the following output:
-- Found Git: /usr/bin/git (found version "2.17.1")
-- Revision status: Detection failed
[ 0%] Built target GitRevision
[ 63%] Built target precice
[ 64%] Built target binprecice
[100%] Built target testprecice
Install the project...
-- Install configuration: "Debug"
-- Installing: /usr/local/lib/libprecice.so.2.0.2
CMake Error at cmake_install.cmake:53 (file):
file INSTALL cannot copy file
Makefile:150: recipe for target 'install' failed
make: *** [install] Error 1
I guess you have to run with sudo rights:
sudo make install. Did you?
I was not able to follow your advice on OpenMPI. Could you please elaborate a bit more?
Sure. For the communication between coupled solvers, preCICE can use two technical implementations: MPI Ports or TCP/IP sockets. You can choose in the
precice-config.xml which one you want to use. OpenMPI does unfortunately not support MPI Ports as we use them. So when you build preCICE with OpenMPI, you cannot use MPI Ports for communication between coupled solvers
m2n:mpi. Still, we also need MPI for other things, so it is still important that you link against OpenMPI. Long story short, my advice is:
- Don’t worry, just use
m2n:sockets and not
m2n:mpi. Ignore the failing test.
- Or, if you are on a large computing cluster and need the last bit of performance, use any other MPI implementation (e.g. MPICH or Intel).