Does anyone else notice the blatantly obvious system bottleneck in all but one of the tests? All of the results are invalid with the exception of the first one and the last one. The second to last bench is marginally more than the margin of error even with a spot on timedemo.