It's both.
Spec is/was a 5v pullup inside the fan, recommended spec is a 3.3v pullup inside the fan w/5v tolerance.
Hence, you can use either one.
Officially speaking you don't need to source any voltage with a fan controller, just sink it. Convincing a 555 timer to sink-only is a good trick though, and the fans certainly don't care if something else is sourcing the pullup (they might care if you sourced 12v or something of course).
Mike, am I correct in assuming that the scope shot is of a 3phase fan? I'm making that assumption based on their being a neutral.
Neither trace looks especially sine waveish to me, but the bottom one is less square for whatever that is worth. The IC you linked specifically says it doesn't care about input PWM frequency.
Blaylock, Blackjackel, whoever wants an answer to this thread's original question: What do you consider "high CFM", and what do you consider to be "quiet"?
I realize it's a cheeky question at best, but it's also rather crucial.
What I consider to be "quiet" and what Mike considers to be "quiet" are two rather different things, for instance.
Similarly what Mike I consider to be "high CFM" (I think we agree on that one, heh. HIGH LIKE HIGH) and what you consider to be "high CFM" are probably quite different as well.