Originally Posted by Dopedangel
I could be wrong but it could be because the original source of the epub is word html. Word html leaves a lot of crap behind in the html coding and converting those also takes a lot longer than normal epubs. The count pages might be having a similar problem.
Seems like a reasonable assumption. I would have guessed the slowness was in the conversion before counting the words, but she said those books converted as fast, maybe faster than the ones where the count finished quickly. Still your idea makes the most sense to a layman such as myself.
The only thing I know for sure, changing Job priority from Normal to Low allows me the freedom to do whatever I want no matter how much work those calibre-parallel processes are doing.