The author considers how the recent release of OpenMP could finally shove parallel software applications out of the domain of high-performance scientific research and analysis, and onto the desktop. By supporting cross-platform, directive-based programming, the OpenMP application programming interface will simplify the development of parallel applications on shared-memory machines. Among other things, this means independent software vendors and other developers can now create parallel code once and run it on different systems, from two-way NT workstations to Unix-based supercomputers with 128 or more processors. The OpenMP specification is a set of compiler directives and callable runtime library routines that take advantage of shared memory multiprocessor (SMP) Unix and NT environments. This version of the OpenMP specification is Fortran-based, due to its widespread use for developing high-performance parallel scientific codes.