Original subtitles

The CPU, short for Central Processing Unit, is like the brain of the computer, and once

you understand how it works, you'll understand the computer as well.

Let's remove the cover of the CPU and zoom in to see what happens inside.

There are lots of different wires carrying information around the CPU.

This particular CPU is called the 6502 and was used in computers like the Apple II and

the Commodore 64, as well as in the original Nintendo Entertainment System.

This simulation of the 6502 can be found online at visual6502.org.

In every CPU, there is a particular wire that turns on and off at a steady rate to help

keep everything in sync.

That wire is called the clock, and the clock in this simulation is turning on about twice

a second.

Modern CPUs are measured in gigahertz, giga meaning billion and hertz meaning times per

second, so the clock in modern CPUs turns on several billion times per second.

That speed is what allows CPUs to do very complicated things very quickly.

However, what the CPU does during each clock tick is actually very simple and something

we'll look at more in this video.

For now, we'll zoom back out and put the cover back on.

The CPU in your computer might be manufactured by a company like Intel or AMD, but the type

of CPU we're going to look at today is called the Scott CPU.

The Scott CPU doesn't actually exist except as a design in a book called But How Do It

Know by John Scott.

The design of the Scott CPU is copyrighted and is being used in this video with John's

permission.

The book is available at buthowdoitknow.com.

This is a great book that goes through each of the components in the CPU very slowly,

without using a lot of overly technical jargon.

If you've been looking for a book that explains how a computer works, I would highly recommend

this one.

So let's flip the CPU over and look underneath.

You'll see a lot of pins sticking out that allow the CPU to take in information and send

it back out.

The CPU fits into what's known as the motherboard.

The motherboard allows all the components in the computer to connect to each other.

So we'll flip the CPU back over and plug it into the motherboard.

On the right of the motherboard is a place for something called RAM.

RAM is short for Random Access Memory, and it just contains all the data that is being

processed by the CPU.

Let's learn a little bit more about RAM by looking at how the CPU and RAM interact.

For now, we'll remove the wires on the left and move the motherboard over to make room

for the RAM chip.

RAM consists of a list of addresses, and at each of those addresses is a piece of data.

The CPU normally requests and processes each piece of data from RAM in order, one after

the other.

However, if the CPU is instructed to pull data out of order, it can do so.

That is why it's called Random Access Memory.

The data can be accessed randomly if it needs to be, although normally it's accessed in

order.

When the computer first starts running a program, it sends an address to RAM to begin retrieving

that program.

The RAM address just consists of a series of ones and zeros, representing on and off

wires.

RAM doesn't do anything with that address, though, until the CPU also turns on the Set

or the Enable wire.

If the Enable wire is turned on, RAM automatically sends whatever piece of data is at that address

back to the CPU.

That data is then processed by the CPU accordingly.

Once the CPU is finished processing that piece of data, it then sends another address to

RAM, turns on the Enable wire, and gets the next piece of data from RAM.

This process happens over and over again inside the computer.

If the CPU needs to save data to RAM, it outputs an address, outputs some data, and then turns

on the Set wire.

The RAM will then overwrite the data at that address with the new data.

But what is that data inside RAM?

Because it just looks like a bunch of ones and zeros.

Well, it's made up of different things.

Some of the most important pieces of data in RAM are the instructions.

Instructions just tell the CPU to do different things.

There are also numbers inside that data.

These are numbers that you might want to add together, compare, or simply process in some

way.

Another thing, and this is kind of weird, that is in the data is addresses.

At particular memory addresses in RAM, the data itself is also an address.

These addresses can be used for various things.

For instance, if you want to output a number to an external device, you have to know the

address of that particular device.

Do you want to send data to the printer?

Or do you want to send it to the monitor, for instance?

There are also letters stored in RAM.

If you want to show some text on the screen, you would actually store it as a bunch of

ones and zeros in RAM.

Each letter is stored as a particular combination of ones and zeros according to a character

code.

These character codes are arbitrary.

Someone just decided that this is a lowercase a, and this is an uppercase G, for instance.

So that's what's actually in the data inside RAM.

Now let's go back to seeing the data as just a bunch of ones and zeros.

We will now move the RAM chip into the RAM socket on the motherboard.

We will then group the RAM addresses and data together, pull the CPU out of the motherboard,

and look at what's called the instruction set of the CPU.

As we saw earlier, some of the pieces of data in RAM are instructions, and each CPU has

its own set of instructions that it understands.

So there might be a load instruction, which loads a number from RAM into the CPU.

After a couple of these load instructions, there might be an add instruction that adds

these two numbers together.

After an add instruction might be a store instruction, which saves the result of that

addition back out to RAM to be used later on.

There might also be a compare instruction after some load instructions, which compares

two numbers together to see which one is larger or if they are the same.

The compare instruction can be very useful when used in conjunction with what's called

a jump if instruction.

As we saw earlier, the CPU generally requests each piece of data from RAM in order one after

the other.

Sometimes, though, the programmer wants to jump to an out-of-order RAM address to process

some other instructions in memory.

The jump if instruction checks to see if a certain condition is true before it jumps.

It uses the results of the compare instruction to make this decision.

There is also a regular jump instruction that jumps no matter what.

Finally, there is an out instruction and an in instruction.

These will output data to an external device, like a monitor, or input data from an external

device, like a keyboard.

These two instructions are often used in conjunction with an address, like we talked about earlier.

There are some other instructions in the CPU's instruction set, but these are some of the

more commonly seen ones.

So as we saw, the data in RAM consists of things like instructions, numbers, addresses,

and letters.

So let's go through a program that would use this instruction set to play a guessing game.

So it would load a number like 9 into the CPU.

Let's say that the programmer decided that that was the right answer, so he went ahead

and put that number into RAM.

And then comes an in instruction to retrieve the user's guess.

After the in instruction is the address of the keyboard, so we can know where we are

getting the data from.

Next comes a compare instruction that checks to see if those two numbers, the one saved

by the programmer and the one entered by the user, are the same.

Following the compare instruction is a jump if equal instruction, which will jump to another

address in RAM if those two numbers we just talked about are the same.

The jump if equal instruction is immediately followed by a new RAM address.

If the two numbers are the same, the CPU jumps to that new address to begin processing its

next set of instructions from there.

If the two numbers are not the same, then the computer ignores the jump if equal instruction

and the corresponding address and just keeps going.

Following the jump to address comes an out instruction with the address for the monitor,

then the letter capital G, and then below that would be the letters U, E, S, S, space

again, so guess again.

So if the user guesses the wrong number, the program would tell him to guess again, and

then jump back up to the in instruction to retrieve that new guess and then process these

instructions all over again.

By the way, the in and out instructions used here have been simplified somewhat, but you

will find them covered in more detail in the book.

So now, let's briefly take a look inside the CPU itself to see how it would process an

instruction.

As we saw earlier, this is the inside of the 6502 CPU.

Let's take away the 6502 wiring and see what's inside the Scott CPU.

The first component is the control unit, which is kind of like a captain in the army.

It receives its orders from RAM in the form of an instruction, and then breaks that instruction

down into specific commands for the other components.

One of the most important components under the command of the control unit is the arithmetic

logic unit, or ALU for short.

The ALU is what performs all the mathematical operations inside the CPU, such as addition,

subtraction, or even comparison like we saw earlier.

The arithmetic logic unit has two inputs.

We'll label them input A and input B, and assume they are two numbers from some previous

load instructions.

Now we might want to add those two numbers together.

The control unit receives that instruction from RAM and then tells the ALU what type

of operation to perform.

The ALU performs the operation and then outputs the answer.

Sometimes though, depending upon the type of instruction, the output from the ALU can

actually be ignored.

For instance, if you have a compare instruction, the ALU doesn't need to output an answer.

Instead, it just needs to tell the control unit how the two numbers compare to each other.

For this, the ALU uses what are called flags, and they help the control unit decide what

to do when it receives the next instruction, like jump if, which we'll see later.

For now though, let's say that we are working with an instruction that does produce an output.

Where does that output actually go?

Well, the eight wires coming out of the ALU would actually run to what is called a register.

A register is a very simple component whose only job is to store a number temporarily.

Registers act just like RAM, except they are inside the CPU, making them faster and more

useful for storing a number temporarily while an instruction is being processed.

When the ALU sends the output to the register, it won't actually be saved until the control

unit turns on the register's set wire.

The set wire is just like the one we saw earlier for RAM.

When the set wire is turned on, the register saves whatever number is on its input wires.

Once we have the output saved in the register though, how do we get it back out?

Well, when we are ready to move a number out of the register, we need another control wire,

called the enable wire, that also runs from the control unit to the register.

As soon as the control unit turns the enable wire on, the register will output whatever

number is saved inside.

The output wires of the register then connect to what is called the CPU bus.

A bus, as we saw earlier on the motherboard, is simply a group of wires that connect multiple

components inside a computer.

On the bus are some other registers with their own set and enable wires.

These may have numbers from previous instructions already saved inside.

So the control unit will then turn on the set wire of the particular register that it

wants to save that number to, and that number will be saved in that register.

Afterward, the control unit will then turn off the enable wire from the first register

and clear the bus.

The four registers at the top are just used for storing numbers between operations, so

they have output wires that go directly back onto the bus.

So now we have moved a number from one register to another just by turning some wires on and

off.

That's the advantage of the bus, easily moving numbers between components.

The disadvantage of the bus is that you can only have one number on it at a time.

Because of this limitation, the arithmetic logic unit uses a temporary register for input

B. When the control unit is processing an instruction involving the ALU, it will move

one of the inputs to the temporary register.

The temporary register has no need for an enable wire since it only outputs to the ALU

and doesn't conflict with any other registers.

The other input to the ALU comes directly from the bus.

The control unit will enable another register and that number will become input A to the

ALU.

That number stays on the bus until the ALU is finished processing the instruction.

And so now there are two inputs to the ALU and we're ready for the ALU to perform an

operation.

As we saw earlier, the control unit knows what operation to tell the ALU to perform

because of the instruction it receives from RAM.

The instruction itself is in another register called the instruction register.

By the way, the input wires from the bus won't affect this register since the instruction

was already saved in a previous step.

This register, like the temporary register, also has no need for an enable wire since

it just outputs to the control unit.

Based upon this instruction, the control unit then tells the ALU what type of operation

to perform.

So let's say the instruction we're processing is a compare instruction.

With a compare instruction, we're not interested in the number that is output by the ALU.

We only want to know how the two inputs compare to each other.

For that, we use the flags that we talked about earlier.

Each flag is just a wire that turns on or off depending upon whether or not a certain

condition is true.

Inside the Scott CPU, there are four flags and we'll look at two of them now.

The A is larger flag will turn on if input A is larger than input B. If the inputs are

the same, then the equal flag turns on, and if both of these flags are off, that means

input B is larger.

But in this case, the equal flag is on, so that means both inputs are the same.

Once the compare instruction is over, we still need to use the flags for the next instruction,

so we'll save them to a register that only has four inputs and four outputs, one for

each flag.

Once the flags are set into the flags register, the CPU is finished with the compare instruction

and it can then request the next instruction from RAM.

Generally, the next instruction after a compare instruction is a jump if instruction.

This combination of a compare and a jump if instruction, by the way, is very common in

computer programming.

Anytime there is more than one possible path through a program, the computer is using these

instructions to tell it which way to go.

So now that we're finished with the compare instruction and our flags are saved in the

flags register, we need to tell RAM that we're ready for the next piece of data, in this

case the next instruction.

So inside the CPU, another register that is very important is what's called the instruction

address register.

The CPU uses this register to know where the next instruction should come from in RAM.

When the CPU is ready for the next instruction, it enables the instruction address register

onto the bus.

Eventually the instruction address will flow to RAM, but it doesn't get there directly.

There is an intermediary register called the memory address register, whose only job is

to tell RAM what memory address the CPU wants next, since it won't always be an instruction.

Once the instruction address is set into the memory address register, it is automatically

sent to RAM, since the memory address register doesn't have an enable wire.

The control unit then turns on the enable RAM wire, and RAM automatically sends back

the data at that address, which in this case is an instruction.

That instruction is then saved in the instruction register, and the control unit begins processing

it.

In this case, it's a jump if equal instruction, which checks to see if the equal flag is on.

It does that by running one of its wires, and the equal flag wire, into an AND gate.

If both inputs to the AND gate are on, then the output wire turns on as well.

This output wire will then trigger the jump.

That jump eventually retrieves the next piece of data from RAM, which happens to be an address,

and move it into the instruction address register.

When the jump if equal instruction is over, the CPU then processes the instruction at

that new address.

At that new address may be some instructions that output the text, you guessed correctly,

onto the screen, because now we know that the user guessed correctly.

So we'll add the final four wires to our CPU, which are used to control the external devices

like the monitor and the keyboard.

We now have a nearly complete picture of what the Scott CPU looks like.

Data moves around inside the CPU using the bus, and is stored in each register according

to how that data is going to be used.

Each instruction that we have seen can be processed by the Scott CPU in about 6 clock

ticks.

Modern CPUs can process multiple instructions per clock tick, meaning that the computer

you're using to watch this video is likely processing tens or even hundreds of billions

of instructions per second.

That phenomenal speed is accomplished by using more than one of each component, and making

sure that all the components are active as much as possible.

This makes modern CPUs much more complicated than the Scott CPU, but they are still fundamentally

doing the same things as the Scott CPU.

So now let's zoom back out and we can see all the wires that run back out to the pins

on the chip.

On the right are the Set RAM and Enable RAM wires.

On the top are the RAM Address wires.

On the bottom are the Data wires that run to both RAM and the external devices.

And on the left are the Input Output Control wires.

So let's zoom back out to see the rest of the chip, and we'll put the cover back on

the CPU and put it back in the motherboard.

Using the ports on the left, we can now plug in the cables that connect our monitor and

our keyboard.

Each of these ports has an address, and that port address is what the CPU uses with an

IN or an OUT instruction.

That port address, by the way, is sent using the data bus, since the address bus in this

computer is only used for RAM.

So we'll zoom out to see how the motherboard fits inside the computer case.

In the computer case is the last component we'll look at, which is the hard drive.

As soon as the power to the computer is turned off, all the data and RAM is lost, so you

have to have a way to store it more permanently.

For that, we use a hard drive.

Inside the hard drive is a spinning disk covered in tiny magnets with a small metal arm floating

above it.

The arm moves around to the different parts of the disk where different data can be stored

and retrieved.

The disk and the arm generally move very, very quickly, but nowhere near as fast as

the CPU can process data.

For this reason, all the data from the hard drive must first be moved to RAM before it

can be processed.

So, we'll put the hard drive back inside the computer and zoom out.

Here we can see the program we just ran and the message telling the user that he guessed

correctly.

So now you've seen the very basics of how a computer processes information.

You'll find much more about the Scott CPU and the book at the website ButHowDoItKnow.com.

Also there are a few small differences between the book and the video, but those shouldn't

detract from your understanding of either.

You can find a list of these differences in the video description.

Thanks for watching.