Performance Benchmarking With the Austin RCS Benchmark Suite: Beyond empirical verification and validation

The realized performance (error–cost tradeoff) of three computational electromagnetic (CEM) methods, which use parallel algorithms on a supercomputer to predict the radar cross section (RCS) of complex targets, are quantified using the Austin RCS Benchmark Suite. The article demonstrates how modern benchmark suites can be used to evaluate CEM methods empirically and compare their performances objectively.

The Austin RCS Benchmark Suite [1], [2] has recently been populated with 20 carefully selected problem sets that span a wide range in six dimensions of computational difficulty [3].