Although the US government collects massive amounts of SAR imagery, the amount of labeled high-value objects is very limited.  Indeed, for some high-value fleeting targets, the collected imagery is barely sufficient for performance testing of a deep learning ML ATR.  EMSI developed a fast ray tracer that is several orders of magnitude faster than the current gold standard for radar simulations and can accommodate all relevant target aspect angles.  Our simulator models layover and shadows faithfully.  This is significant because Deep Learning ATRs can exploit shadow and layover information.  The backscatter return is added coherently to collected SAR background imagery.  There are many methods for specification of realistic target placement and relationships between associated targets can be programmatically specified.  A radar camouflage model and a computationally efficient ground bounce model are available.  Its full multistatic capablility enables its use for areas of denied access.  The simulator uses facet target models that are available from government sources or video gaming suppliers.