Low-level systems stuff, and when dealing with file-formats.

For day-to-day stuff you've got int, float etc

This is one of the reasons why we have to have so much memory now days.

If I've got a variable that has a range of 0-9001, I'd likely still put it in and [font="Lucida Console"]int[/font] rather than a [font="Lucida Console"]short[/font], unless there's a good reason to reduce the memory usage (e.g. serialisation across network/disk/system-bus). Normally you just want to stick with the CPU word size to avoid packing overheads.
[/quote]
Indeed, there can be some reasons to use larger variables than needed (alhough reducing memory usage is not the only reason to avoid them; think of a video player, even a 1920p video frame isn't big for today standards. Would you dare to work with only ints to, say, adjust contrast and brightness at a low level?)

Ignoring the fact that you are oversimplifying, this is a discussion in a beginner's thread where some dude is asking whether he must learn the different basic types of his choice.
I find shocking me that the message most of you are transmitting is: "D'oh, who cares".
Yeah, perhaps most of the time you can just throw an int. Perhaps most of the time it's even better. But you do it for a reason, not because you don't have another choice. In the same way that TheTroll chooses to use the smallest possible data type. You can argue against the wisdom on that, nowadays that we count RAM in Gb and hard discs in TB, but the fact is that both of you know what you need, balance the advantages and the disadvantages, and make your choice, be it better or worse.

GraySnakeGenocide, you can probably do almost the same things without knowing those data types (not everything, at least not in a overly complex way). However those types are there for a reason. I personally don't find very compelling to work with pro's who can't say the max value of a 32 bit integer (I don't need the exact value, but should know it's around 2000 millions if it's signed).
Probably knowing the exact values is overkill, but you should get a grasp of them, and know the reason (signed: -2^(n-1) to (2^(n-1) - 1); unsigned: 0 to ((2^n) - 1).
You should know that 32767 + 1 = -32768 when you are using ushorts, and that 65535 becomes -1 when converting a short to an ushort. And, most important, you should know WHY.

The time you spend learning this will be saved in frustrating clueless debugging sessions

Dude, I feel so 90-ish now. I'll better take a shower.