MPI-CUDA parallelization of a finite-strip program for geometric nonlinear analysis: A hybrid approach