**4-bitCPU ALU Adder Simulator**

Hello! This shows how computers do integer (whole-number) addition in binary. The simulator above lets you tinker around with addition and
subtraction like a computer would see it. Being interactive is supposed to help the process feel more intuitive, so **go ahead and play with it**
and read on if you want more details.

This tool is a companion to my **Sprout24 CPU** project. I'm building up some documentation and learning material, and this
is going to be part of that larger effort.

**Binary Digits and Unsigned Numbers**

We are looking at a **4-bit adder** here, which means that as far as this simulated computer can represent,
there are **2 ^{4}** â or

**16**â different possible numbers.

Since binary is another word for Base-2, each digit can only have two possible values. In our usual Base-10, each digit can be
any number 0â9. Here, though, **each digit can be only be 0 or 1**. "Off" and "on" are another way to think about it.

In this simulator, "1" or "on" bits are highlighted red, while "0" or "off" bits are grey. See the top two rows of bits labeled "**Input A**"
and "**Input B**"? **Go ahead and click on any of those input bits.** By switching them on and off, you're changing the digits of the two binary numbers
that the computer is going to be adding together.

The most natural way to understand these numbers is that any combination of 4 bits will map to numbers from
zero (all bits off) to 2^{4} - 1. Those are considered **unsigned numbers: 0 through 15**.

**Negative Numbers and Ones-Complement**

That's simple enough. But **negative numbers exist, too**, and there's no room for them in that range. What if we could split that range in
half? We could leave half of the numbers on the positive side, but let the other half represent negative numbers.

One way to do this is to just reserve the top bit to indicate whether the rest of the number is positive or negative. So going from
all-bits-off to all bits-on, you'd count up from **zero up to 7**, then flip the sign and keep going from
**-7 down to -0** before running out of digits.

Fair enough. That representation for signed numbers is called **ones'-complement**, and some old computers did use it. But **it's kind of
wonky**. We'd have two sort-of-equivalent numbers with different representations: **+0** (all bits off) and **-0** (all bits on). And it turns out there's a
better way to handle it.

**Twos'-Complement: Better Negative Numbers**

What if we shift the representation to get rid of one of those redundant zeroes? Instead of -0, what if we decide that all-bits-on represents -1
instead? Then when we go from all-bits-off to all-bits-on, we go **zero through 7**, then the next
number is **-8** instead of -7, then count down
the rest of the way until **all-bits-on is -1**. This is called a **twos'-complement representation, and is what basically all modern computers**
use, including the Sprout24 CPU and **this simulator**.

And it ultimately makes the math and the design of the computer easier. **Changing the sign of a number requires flipping all the bits and adding 1**.
But when you do that, subtraction works out more cleanly.

**Subtraction**

Since **subtraction is just the inverse of addition**, we can **turn any subtraction operation into an addition operation** by switching
the second number from positive to negative (or vice-versa!) and just adding those two numbers. So in the simulator here, notice that if you
click the "**Inv (Subtract)**" checkbox, the value in the "Adder B" box changes. Because we're using twos'-complement, it's just all the bits from
"Input B" flipped, then adding 1 to that modified value. From then on, the adder just treats it as any other addition.

Note that **"Adder A" is always the same as "Input A"**. It's only the **second number "Input B" that gets negated** when we're treating the
operation as a subtraction.

**Signed vs. Unsigned**

This all means that the **same combination of 4 bits** might have a **different meaning** depending on whether we're
interpreting those bits as an unsigned or a signed value. For the first half of the range, numbers mean the same thing in unsigned and signed versions.
It's only when you get to the top half that they diverge in meaning.

This simulator shows every number in three columns on the left. The **first column is the bit pattern interpreted as a signed number**.
The **second column interprets it as an unsigned number**. And the **third column** is the same unsigned number, just in **hexadecimal**
(Base-16) instead of Base-10.

**Input Fields**

In addition to clicking on the input bits to change them, you can just enter numbers into the two fields on the right side. The field always shows the unsigned interpretation of the numbers, but you can type in a valid negative number and it will auto-convert for you. Just hit enter after putting in a number, and it will run the addition and recalculate everything.

**Carries**

The carry indicators above the two Adder values light up when the column to its right has two or more of its A, B, and Carry bits lit up. It's just like carrying the 1 when doing regular old Base-10 addition, but in this case the carry happens when the column adds up to 2 or more rather than 10 or more.

You can see how it takes time for the result and carry bits to propagate from right to left, since each column depends on the column to the
right, and **the speed of transistors is not infinite**. In this simulator, we've slowed it down to about 50 milliseconds per bit, but in a real
computer we're measuring in nanoseconds. Still, the propagation delay is actually enough of a concern that real computers (including the
Sprout 24) use some tricks to combine some pieces of the carry operations to "look ahead" and allow the addition to complete faster. Since the
**CPU can only clock as fast as its slowest propagation path**, it makes sense to add the extra complexity and logic gates to try to avoid long
propagation times. This simulator shows the basic propagation method since it's easiest to understand, but I may add an option to visualize
the carry-lookahead method as well in the future.

There's **one carry bit past the highest column** at the top that is generated by the result of that highest column. We call that the
"carry out" value, and we keep track of it so that the program running on the computer can tell the answer is large enough that it doesn't
fit in the result value.

**Flags**

With that, let's look at the "Flags" box in the lower-right corner. Those are four signals that the adder sends back to the rest of the CPU along with the addition result. Here's what they mean:

**(Z) Zero**: This signal is turned on when **all the bits in the result are off**, meaning the **result of the operation was zero**.

**(S) Sign**: This signal is turned on when the **most-significant bit of the result** (position 3) is **turned on**.
This is an indication that, **if you're interpreting the result as a signed number**, then you need to **view the result as negative**.

**(C) Carry**: This signal is turned on when the result of **adding two unsigned numbers gives a result larger than what fits in the result**,
and the top bit "spills out" the top through the Carry-Out bit. When the checkbox for Subtraction mode is checked, this value is inverted
so that it is the opposite of the Carry-Out bit, which in subtraction effectively means the Adder B value was larger than Adder A and it
would have had to "borrow" a bit from the next column if it could. Note that the carry flag **only makes sense when interpreting values as
unsigned**; it's not meaningful when interpreting numbers as signed values.

**(O) Overflow**: This serves a similar purpose to the Carry flag, but **for operations interpreted as using signed numbers**. Basically if you
add two positive signed numbers and the result is negative, or you add two negative numbers and the result is positive, then you know your **addition
spilled over across the sign boundary** and the the result isn't what it seems. The Overflow flag is meaningless if you're interpreting the values
as unsigned numbers.

These flags are important because they allow the CPU to compare values and effectively make if-then decisions. That's a topic for another day.

**What Else?**

This simulator was built to show 24-bit addition to match the Sprout24 CPU, but it's easier to intuitively grasp all this with smaller ranges of numbers. If you want, you can switch to 8-bit or even 4-bit modes to be more cozy. Remember that our modern computers are mostly 64-bit, so the numbers they work with natively can be galactic-scale enormous.

This page is a draft for a larger set of learning material, but it's useful enough that I wanted to share it now. I looked for a similar interactive adder out there and came up pretty empty, so hopefully this proves helpful. Thanks for checking it out.

~ Autumn

@[email protected]