Numerical experiments performed on a single instruction multiple data-pipeline vector parallel (SIMD-PVP) architecture computing machine, e.g., a CRAY X-MP/48, demonstrate that current nuclear reactor systems codes can be restructured for concurrent multiprocessing and show wall clock performance improvements of 1.5 to 3.0 on a 4-CPU machine, depending on plant model, problem type, and problem length. In addition, algorithm development studies indicate that up to a 20% speedup can be obtained by a new class of parallel numerical methods. Faster-than-real-time simulation has been demonstrated utilizing RELAP5/MOD1 and a pressurized water reactor plant model characteristic of licensing and/or safety analysis calculations. A theoretical analysis indicates that five to ten times faster than real-time computation may be possible for this class of problems utilizing this or the next generation of SIMD-PVP architecture machines, such as the CRAY X-MP/48, and new computer codes optimized for such machines.