We get segfault running the brusselator on Summit early in the run process with gcc/6.4.0. This appears to be happening in the heat2d_fft code, in the call to MPI_CART_SUB. I can reproduce this with a vanilla installation of 2dcomp_fft with the gnu compiler (but not with xl.) It looks like the MPI call is being redirected to a PMPI call, and I suspect something is broken on the installation there. I have a minimal example that reproduces the issue, and I am submitting an OLCF ticket.
We get segfault running the brusselator on Summit early in the run process with gcc/6.4.0. This appears to be happening in the heat2d_fft code, in the call to MPI_CART_SUB. I can reproduce this with a vanilla installation of 2dcomp_fft with the gnu compiler (but not with xl.) It looks like the MPI call is being redirected to a PMPI call, and I suspect something is broken on the installation there. I have a minimal example that reproduces the issue, and I am submitting an OLCF ticket.