Data Representation

How Do Computers Store Data?

●
Remember when we said that it all comes down to 0s and 1s?
●
This is how data is stored on computers
●
These are really voltages (low for 0 and high for 1)
●
Storage Units
– Basic Storage Unit: Binary Digit (Bit)
– Basic Grouping: 8-bits (Byte)
– Natural Unit of Data: Word (Varies by computer – example 64-bit
computer)
 Data Representation
●
In computers, data is represented in many different ways:
– Numbers
– Text
– Audio
– Images
– Video
●
All of these are made up of bits
 Analog vs. Digital Data
●
Analog is continuous data, for example the range of real
number values from 0 to 1
●
Digital is a discrete (countable) representation
– The data has gaps between elements
●
Digital computers are by far the most commonly used today
●
Although analog computers fell out of favor, in recent years,
they are making (a bit of) a comeback
 Data Representation: Numbers
●
We have seen how non-negative integers are
represented using bits (binary numbers)
●
What about negative integers?
●
What about real numbers?
 Data Representation: Numbers
●
If integers are infinite, how do we fit them into a
finite amount memory?
– There are limitations to the maximum value
– Trade-offs
●
Accuracy
●
Maximum value
 Data Representation: Numbers
Let’s look at Positive Numbers in 8-bits of Memory

When Integers are stored as only Positive or 0, we call them:

Unsigned Integers
 Negative Numbers
●
We have looked at positive integers in binary
●
What about negative values?
– These examples will use 8-bits for the numbers but
the concepts translate to any size of word
 Negative Integers in Binary
●
Binary numbers use a single bit to state whether the number
is positive or negative
●
This bit is called the “sign bit” and is found in the leftmost
place of the number
The sign bit goes here -> __ 0 1 1 0 1 1 1
●
If the sign bit is 0, the number is positive
●
If the sign bit is 1, the number is negative
●
You need to explicitly state if an integer is signed or unsigned
 Negative Integers in Binary
●
We will look at three different ways negative
integers are represented in binary
1) Signed Magnitude
2) One’s Complement
3) Two’s Complement
 Signed Magnitude
●
Also called “Sign Magnitude” and “Sign and Magnitude”
●
Just use the most significant bit (left-most) as the sign bit,
and the remaining 7 bits for the conversion:
011000112 = 9910
111000112 = -9910
●
If we were using unsigned integers then all 8 bits are used
for the conversion:
111000112 = 22710
 Signed Magnitude
Example
If this number was unsigned it would be 13310
10000101

If it’s signed? Well, now the left-most 1 represents the sign.

Skipping the left-most 1, we get 0000101, which is 5.
So, in sign magnitude notation, 10000101 would be -5.
 Signed Magnitude
●
So, what would this number be?
00000000
●
Yes, it is of course, zero. And what would this number be?
10000000
●
This is negative zero. Is zero a positive or negative integer?
●
Having +0 and -0 makes math difficult, and makes
computers less efficient
●
Let’s look at another method
 One’s Complement
●
To get the negative value of a number (i.e. multiply the value by -1), we
flip all the bits. All 0 bits become 1 bits and all 1 bits become 0 bits.
011000112 = 9910
100111002 = -9910
●
Complements have interesting mathematical properties (which are
beyond the scope of this discussion)
●
However, the addition is still not completely easy, and there’s still the
problem of two 0s:
000000002 = +010
111111112 = -010
 Two’s Complement
●
Two’s Complement is what is used on virtually all computers today.
●
To get the negative value of a number there are two steps:
Step 1: Flip all the bits (just like with One’s Complement)
011000112 = 9910
100111002
Step 2: Add 1 to the result
100111002
+1
10011101 = -9910
 Two’s Complement
●
Two’s Complement gets rid of the two versions of zero. Let’s try to get -0
Flip the Bits: 00000000 → 11111111
Add 1 to the result:
11111111
+1
100000000
●
Note that there are 9 bits in the result. The leftmost bit is a carry bit that can’t
be held within 8 bits. So, it is discarded, leaving the result as:
000000002 = 010
So, only one representation of 0
 Two’s Complement - Math
●
But, does math work properly?
●
It turns out yes. Two’s Complement has a
straightforward way to do math
●
Suppose we want to subtract 1 from 10
10 - 1
●
Instead of subtracting 1, we add its complement
10 + (-1)
 Two’s Complement - Math
000010102 (1010)

+ 111111112 (-1 in Two’s Complement)

1000010012 (result in Two’s Complement)

●
Note that we have an overflow bit. Truncating it
gives us 00001001 which is 9
●
In Two’s Complement, all you need to do is add
 Two’s Complement – Find Positive
●
We know how to get the complement (negative value) of a positive
integer. How do we get the positive value of a negative integer?
●
No need to reverse the process! Just follow the same process:
Step 1: Flip all the bits (just like with One’s Complement)
100111012 = -9910
011000102
Step 2: Add 1 to the result
011000102
+1
011000112 = 9910
 Practice
● What is the negative representation of 4510 (001011012) in:
– Signed Magnitude:
101011012
– One’s Complement
110100102
– Two’s Complement
110100112
 Practice
●
Give the binary (using 8 bits) representation of
the number -118 in
– Signed Magnitude
– One’s Complement
– Twos Complement
 Practice Answers
●
Give the binary (using 8 bits) representation of the
number -118 in
●
118 in binary is 01110110
– Signed Magnitude
●
Just flip the sign bit --> 11110110
– One’s Complement
●
Flip all the bits --> 10001001
– Twos Complement
●
Flip all the bits and add 1 --> 10001001 + 1 = 10001010
 More Practice
Show the steps to do the following operation using two’s
complement:
74 – 83 = ?
First, find the binary values of the numbers 74 and 83:
7410 = 010010102
8310 = 010100112
Next, take the two’s complement of 83:
-83 = 10101101
 74 – 83 = ?
Now, add -83 to 74:
01001010 = 74
+ 10101101 = -83 in two’s complement
11110111 = a negative value (sign bit is 1)
The result is negative, but what is its value?
Find the two’s complement of 11110111
00001000 Flip the bits
+ 1 Add 1
00001001 = 9
So, the result of 74 – 83 is -9
 Two’s Complement - Overflow
●
When adding/subtracting using two’s
complement, if one value is the opposite sign of
the other, overflow will never occur.
●
When overflow does occur, the carry bit is
thrown away.
●
What happens to results if the carry bit is
thrown away?
 Two’s Complement - Overflow
What is the result of 118 + 31 using 8-bit two’s
complement?
11810 = 011101102
+ 3110 = + 000111112
14910 = 100101012 <- A negative value

Programmers must be mindful of the size of the integer

values that they are working with to avoid integer overflow
 Floating Point Numbers
How can we represent real numbers based on what we know about integers?

We just write things in normal base 10 as always.

internal binary representation: like scientific notation.

sign exponent fraction

S: sign bit E: bias-adjusted exponent M: adjusted fractional value
value = (-1)S * 2E * M
Java float: 32-bit representation. (1 sign, 8 exp, 23 frac)
Java double: 64-bit representation. (1 sign, 11 exp, 52 frac)