When calculating orbital paths for unmanned spacecraft how does NASA round off its numbers when doing calculations if at all?
Do they use unrounded off numbers throughout the math and round it off at the end? or never round off?
thanks.

In what context, specifically? What are you attempting to understand?

Well like when they plotted the path for voyager to swing around planets using gravity assist etc they would I think calculate gravity of the planets and elliptical orbital paths of spacecraft etc and had to plot that all out. Lots of simulators like celestia etc to must round off numbers sometimes to mimic accurately what paths historical craft have taken? I am not sure even if they round off that is why I am asking. I would think you would need super precise numbers to make sure everything was timeout perfectly for the route.
or maybe a part 2 to the question is how do solar system simulators get so accurate? If so how big of numbers do programmers need to create an accurate simulator? Can 3d software programs out their work as well as simulators?

This is more of a computer science question than a specific spaceflight question. You can devote as many bytes as you want to store a value to improve its precision, but of course there's drawbacks.
I'd suggest you look at this link to get some ideas.
http://en.wikipedia.org/wiki/Double-precis...ng-point_format

thanks..that info is propabably beyond my capablilites....
but I suppose that was the ultimate question for me was how precise does plotting a trip to the planets have to be for Nasa? Maybe just not a simple answer like I was hoping.

An appropriate introduction might be the documents regarding Deep Space 1 and MER Navigation at the Descanso page here...

http://descanso.jpl.nasa.gov/DPSummary/sum...orce_external=0