We developed a data parallel code written in High Performance Fortran for simulating magnetized fluid flow in thermonuclear fusion devices and solar coronal loops. We compare execution times on a variety of architectures for physically realistic applications. In particular, we quote times measured for high resolution runs on a vector (Cray C90) and several distributed memory parallel architectures (cluster of workstations, CM-5, IBM SP2”, Cray T3D, Cray T3E). On all parallel platforms, we demonstrate almost linear speedup.