Phase I of the SEAM Project will produce a variable-density acoustic synthetic survey over a 3D geological model simulating a deepwater subsalt exploration target. Due to the intended use of the data, the project places a premium on accuracy. Commercially produced Phase I synthetics will be spot-checked against benchmark simulations to assure quality. Thus the accuracy of the benchmark simulator required careful assessment. The authors designed and implemented the benchmark simulator used in this program, subjected it to verification tests, and assisted in the qualification phase of the Phase I project. The key lessons that we have learned so far from this assessment are that (1) the few verification tools available to us - a few analytic solutions and Richardson extrapolation - seem to be adequate, at least in a rough way, and (2) the standard approach to this type of simulation - finite difference methods on regular grids - requires surprisingly fine grid steps to produce small relative RMS errors for models of the type defined by this project.