The multi-core performance of a legacy Fortran benchmark unsuited to data parallelism is enhanced by threading using the TASK construct of OpenMP 3.0 and the Intel Fortran Compiler. The necessary source code changes are explained in detail.