DFT runs stop with “programming error in distribution of points in parallel DFT” with 64bit integer Intel MPI¶
This is not a programming error in DIRAC but a bug in 64bit integer Intel MPI
mpi_reduce. The Intel developers are informed and working on a patch. In the
meantime, for DFT calculations using Intel MPI you should use 32bit integers.