The Speediest results are momentary and depending on a plethora of different factors that can change literally from second to second. To get anything resembling a definitive test you would need to:
  1. take several speed tests taken one after the other. (I would a bare minimum of 5 trials)
  2. repeat that suite using different server locations (By the way the more distant severs may actually result in higher speeds than a relatively nearby server — or not, again I would use 5 widely located target servers)
  3. repeat both suites tests at least five times during a 24 hour day
  4. for more accuracy repeat the entire set of tests over several days of the week
  5. Finally take the average of all of those results.
Of course there will also be monthly and seasonal differences as well, but that should derive a repeatable result with a reasonable statistical margin of error.

If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein