OK, just tested and it seems you're correct. The difference between 15 and 16 OCs over a stack of cobble was noticeable but I couldn't determine a difference between 16 and 17. Even a 0.3 tick per operation when multiplied 64 times becomes almost a second which should've been noticeable even with my Mk I Optical Measuring Device. We can thus conclude that the 1.3 is rounded down to 1 which adjusts my numbers by 1, as you say.
Thanks again for your feedback.
And to any of you who haven't been to university, this is what's called Peer Review and it's a foundation of modern science. SCIENCE ROCKS!
Indeed, and to science! 
Yeah, the code doesn't allow for partial tick operations, it rounds to whole numbers. And yes, my tests concluded the same thing (and I see you've updated the wiki). So grats we figured it out 
[I tried with 1 less and 1 more overclocker .. along w/ the online stopwatch and it's ~3 seconds vs ~6 seconds for a stack (when it takes 2 ticks / operation) so it is easy enough to notice... and was amusing to assemble... and I only blew up 1 furance by forgetting to put in the transformers before hooking up to power :)]
Now Metalchic just needs to figure out how to supply ~3700 EU/t to a Macerator w/ 16 overclockers...