Link ads can be surprisingly good if there's a history of them returning highly relevant links.
As for the CPM, it's about finding stuff that works for your layout and your audience. There's no one size fits all answer. I've lost count of how many experiments I've done over the last 13 years. (At the same time, I'm a rank amateur at testing because I'm rubbish at web design.)
One thing to consider: removing ads can sometimes make you more money. (It's impossible to know without testing.) And why? Because Google attempts to fill all the ad slots on a page. If there's one ad slot, it will automatically hold the highest paying ads Google can find. If there are 3, Google will try and optimise them by moving ads around so that the better ones are in the slot where people seem to click more, but you'll still have 2/3 of ads that are lower paying (by definition: you've got 3 lots of ads, not 1 lot of ads) so if the extra ad units end up spreading the clicks around, your overall CPM could actually drop.
I know people who make many times my CPM, and I know people who make less than £1 CPM and still make a great living from it (the topic's popular so the traffic is high, but it's not something that lends itself well to advertising).
Anyway, there are loads and loads and loads of articles out there about optimising Google ads in all sorts of ways. I won't even pretend to have the "right" answer. The only advice I will give is this: test, test, and test again. If something works, stop (for a while). But be prepared to start testing all over again as stuff changes (like the rise of tablets, and mobile).
.