This should go without saying, but benchmarking to test the raw performance of an engineering sample is a fools errand. Generally engineering samples are carefully selected for stability and testing features, not for their performance or efficiency.