Comment 15 for bug 913615

So it is the same thing that I'm seeing then and no extra factors like completely different filesystem (ext3 vs ext4 is clone enough), good.

The (dramatic) slowdowndown is a somewhat peculiar combination of rpmdb size, system RAM and cache size: with "average" rpmdb sizes things are ok with the current 64Mb cache (this value was the same throughout F15) even with lower RAM, so for example your laptop's own db didn't exhibit the slowness but the desktop systems larger db crawled on the laptop.

Rebuilding the db is a special case though and doesn't really benefit from caching, whereas larger (than 1Mb) cache IS beneficial for other functionality. So its a question of finding a sweet spot that doesn't kill lower end systems while having a reasonable cache on "normal" operations...
How much memory do your systems have, circa 1GB or so?