File it under "register"

That's the best I could come up with, sorry. I'm talking about the title of this post. Sorry.

If you couldn't guess by that awesome title, this post is about the register file. Go ahead and read the title again. Get it? It was pretty good after all, I guess. Anyway, the register file is a part of a computer processor. The simplest way to describe what it is is to say that it's where the variables of a program live when they're not sitting in the computer's memory. The register file is the most temporary of all temporary storage. It is comprised of several "registers," which are just places in the processor that can store a number. If you have a 32-bit processor, then each of the registers in the processor can hold a 32-bit number. It is common for a processor to have from 8-32 (or more) registers.

Data is moved into the registers from memory, some kind of mathematical operation is done with that data, and then the program moves on, putting new data into the register file (the contents of the register file at any one moment doesn't stay relevant for very long usually). If we still need the old data (or the result of some mathematical operation), then we can store the contents of a register back to the memory.

Registers are parts of the computer that can store information, just like memory, however the registers do not have addresses like the memory does. Well, they don't have addresses in the same way, at least, but they do still have a way of telling them apart. Each register has its own name, which is commonly something like $r0 or $r3. The '$' is just convention for saying that we're talking about registers, and then the 'r' is just a further reminder. A 32-register processor would have registers with names from $r0-$r31.

If you'll remember when we talked about branch instructions, all of the branches depended on comparing two different variables to each other, and depending on the comparison, the branch is taken or not. Well, in a real computer we can't just talk about variables as abstractly as that. We have to be more concrete. We have to use registers. A real "branch when x is greater than y" instruction would really look more like this:

bgt $r3, $r4

It's up to the programmer to remember that $r3 really means 'x' and that $r4 really means 'y.'

Load and store instructions are the same way. We don't store 'x' into a memory location, and we don't load into another variable 'y.' Loads and stores look more like this:

ld $r3, 4008 (ld is short for "load")

st 4012, $r4 (st is short for "store")

Every time that you perform some action in a processor, it will involve registers. Are you adding two variables and saving the result in another variable? You're really talking about registers there.

You might think that 8 or 32 registers isn't enough to store all the variables of a program (especially really huge programs like web browsers or word processors). You're right, it's not enough. Most of the variables of a program sit in the computer's memory most of the time. They are only temporarily brought into the registers when they are needed, and then stored back into the memory when they're done. This is happening millions and billions of times every second in your computer right now.

Next time we'll talk about more mathematical operations we can do with binary numbers, but we still aren't going to talk about negative numbers yet. I need to gear up for that one.