What would be much more interesting than "peak responses per second," which is a weird metric to begin with, is the actual histogram of sampled throughputs. Or at the very least a box-and-whisker plot (http://en.wikipedia.org/wiki/Box_plot).
Most folks who have run Rails at scale, for example, find that the untuned garbage collector in MRI (Ruby's default interpreter) introduces a large amount of variance, for example.
We don't have a box plot, but we have line charts for all of the data (performance at multiple client concurrency levels). Just click the "All samples (line chart)" tab on the data panels.
(Note that the very first data panel in the intro is an image and doesn't have tabs.)
Most folks who have run Rails at scale, for example, find that the untuned garbage collector in MRI (Ruby's default interpreter) introduces a large amount of variance, for example.