Parallel computing of groundwater models, i.e., the simultaneous utilization of processors when solving numerical problems, can reduce run times and subsequently reduce cost. Yet not all groundwater modeling simulators that can accommodate parallel computing are designed to do so efficiently. Only few simulators were designed with high-performance computing in mind, where hundreds or thousands of processors can be used effectively at one time (so-called “embarrassingly” or “massively” parallel simulators).
At some point, scalability – i.e., reduction of run time as a function of adding more processors – becomes less effective due to slow-down caused by passing of distributed information from the various processors and “stitching” all this information back together. The ability of a numerical simulator to scale up affects the choice of hardware or cloud computing configuration.
Click here to read the full article
