I haven't seen a supported (publically documented) option for this specific purpose. #pragma distribute point may have such an effect, probably between the loops you don't want swapped. #pragma omp often suppresses loop interchanges.
For more complete information about compiler optimizations, see our Optimization Notice.