I've pushed it around on a few topics, but perhaps this analogy could shed light on why having a lot of files kills your transfer speed as opposed to one big file of the same size.
You're sorting files for a company, you got all the cabinet space you need, but because it's a big company, they're pretty anal about how to index things. Let's say at the minimum, you have to index by Department -> Project -> Date. And these files could be anything from memos, reports, notes, etc. Now you're given two tasks: organize 10,000 separate, distinct sheets of paper; or organize 34 300 page (10,200 pages total) notebooks. Which one do you think will take you less time?
Even though it's roughly the same amount of data, one will take longer because there's more overhead.