Computer Architecture 2023-24

(WBCS010-05)
Lecture 2: Number Systems

Reza Hassanpour
[email protected]
Introduction
› Computers work with digital signals, which
are discrete and have only two possible states:
0 or 1.
› Binary numbers are the most natural way to
represent these two states.
› The binary system simplifies the
representation of digital data by aligning with
the on/off states of electronic switches.
› Question:
• How can we represent decimal numbers in
binary?
Positional and Non-Positional Number
Systems
› Positional number system is the type of number
system in which the value of a digit depends upon its
position in the number.
• Ex. Decimal Numbers

› In non-positional number systems, any symbol has a

specific value regardless of its position.
• Ex. Roman Numbers I, II, III, XX, XXV
Decimal Number System
› The decimal number system is a positional number
system.
• Example:
3 7 1 1 1 X 100 = 1
103 102 101 100 1 X 101 = 10
7 X 102 = 700
3 X 103 = 3000
 Decimal Number System

› The decimal number system is also known as base 10.

› The values of the positions are calculated by taking 10

to some power.

› It uses 10 digits, the digits 0 through 9.

Binary Number System
› The binary number system is also known as base 2.
The values of the positions are calculated by taking 2
to some power.
› Why is the base 2 for binary numbers?
› Because we use 2 digits, the digits 0 and 1.
Binary Number System
› The binary number system is also a positional
numbering system.
› Instead of using ten digits, 0 - 9, the binary system
uses only two digits, 0 and 1.
› Example of a binary number and the values of the
positions:
1 0 0 1 1 0 1
26 25 24 23 22 21 20
Conversion From Decimal To Binary
› Use repeated division by radix.
› Example: 56(10)

› Collect remainders in reverse order

› 56(10) = 1110002
Conversion From Binary To Decimal
› Multiply each digit by radix to the power of the
position of the digit.
› Add up to get the decimal number

Conversion from binary to decimal

1012= 1×22 + 0×21 + 1×20 = 510
Signed Integers

› How to represent positive (+) and negative (−)?

› With n bits, we have 2n encodings
• Use half of the patterns (the ones starting with zero) to
represent positive numbers
• Use the other half (the ones starting with one) to
represent negative numbers
› Three different encoding schemes:
• Signed-magnitude
• 1's complement
• 2's complement

10
Sign-Magnitude
› Uses one bit to represent the sign
0 = positive, 1 = negative
› Remaining bits are used to represent the magnitude
› Range - (2n-1 – 1) to 2n-1 - 1
where n=number of digits
› Example: Let n=4:
› Range is –7 to 7 or
1111 to 0111
1's complement

› To get a negative number, start with a positive number (with

zero as the leftmost bit) and flip all the bits -- from 0 to 1,
from 1 to 0

› Examples -- 5-bit 1's complement integers:

12
Signed-Magnitude and 1's Complement Disadvantages

› In both representations, two different representations of zero

• Signed-magnitude: 00000 = 0 and 10000 = 0
• 1's complement: 00000 = 0 and 11111 = 0

› Operations are not simple

• Think about how to add +2 and −3.
• Actions are different for two positive integers, two negative
integers, one positive and one negative

› A simpler scheme: 2's complement

13
2's Complement

› To simplify circuits, we want all operations on integers to use binary

arithmetic, regardless of whether the integers are positive or negative
› When we add +X to −X we want to get zero
› Therefore, we use "normal" unsigned binary to represent +X.
And we assign to −X the pattern of bits that will make X + (−X) = 0

› NOTE: When we do this, we get a carry-out bit of 1, which is ignored

› We only have 5 bits, so the carry-out bit is ignored and we still have 5 bits

14
 Unsigned

2's Complement Integers Representations

1110
1111
15
0000

0
0 0001
1
0010
14 -1 +1
2
-2 +2
4-bit 2's value 4-bit 2's value 1101
13 0011
-3 +3 3
complement complement Signed Values
(2’s Complement) +4 4 0100
1100 12 -4

0000 0
1011
11
-5

-6
-+ +6
+5
5
0101

0001 1 1111 −1 10
-7
-8
+7 6
0110
1010 7
9
0010 2 1110 −2 1001
8
1000
0111

0011 3 1101 −3
0100 4 1100 −4
0101 5 1011 −5
0110 6 1010 −6
0111 7 1001 −7
1000 −8

› With n bits, represent values from −2n −1 to + 2n −1 − 1

› NOTE: All positive numbers start with 0, all negative numbers start with 1
15
Signed Numbers: 4-bit Example
Decimal 2’s comp Sign-Mag
-8 1000 N/A
-7 1001 1111
-6 1010 1110
-5 1011 1101
-4 1100 1100
-3 1101 1011
-2 1110 1010
-1 1111 1001
-0 0000 (= +0) 1000
16
Converting X to −X

1. Flip all the bits (Same as 1's complement)

2. Add +1 to the result

› A shortcut method:
› Copy bits from right to left up to (and including) the first '1'.
Then flip remaining bits to the left

17
Converting Binary (2's C) to Decimal

1. If leading bit is one, take two’s complement to get a positive number

2. Add powers of 2 that have “1” in the corresponding bit positions

3. If original number was negative, add a minus sign n 2n

0 1
1 2

X = 01101000 two
2 4
3 8
4 16
= 2 + 2 + 2 = 64 + 32 + 8
6 5 3
5 32

=104 ten 6
7
64
128
› In the
8
following 256
table, read
‘2n’ as 2 to
9 512
› Examples use 8-bit 2’s complement numbers the power
n.

10 1,024

18
More Examples
X = 00100111two
n 2n
= 2 + 2 + 2 + 2 = 32 + 4 + 2 + 1
5 2 1 0 0 1
1 2
= 39ten 2 4
3 8
X = 11100110 two 4 16
5 32
− X = 00011010 6 64

= 24 + 23 + 21 = 16 + 8 + 2 7
8
128
256
= 26ten 9 512
10 1,024
X = −26 ten
› Examples use 8-bit 2’s complement numbers

19
Converting Decimal to Binary (2's C)
› First Method: Division
1. Find magnitude of decimal number (conversion always positive)
2. Divide by two – remainder is least significant bit
3. Keep dividing by two until answer is zero, writing remainders
from right to left
4. Append a zero as the MS bit; if original number was negative, flip
bits and add +1
X = 104 ten 104 / 2 = 52 r 0 bit 0
52 / 2 = 26 r 0 bit 1
26 / 2 = 13 r 0 bit 2
13 / 2 = 6 r1 bit 3
6 / 2 = 3 r0 bit 4
3 / 2 = 1 r1 bit 5
X = 01101000 two 1/ 2 = 0 r1 bit 6
20
Converting Decimal to Binary (2's C)
n 2n
› Second Method: Subtract Powers of Two 0 1

1. Find magnitude of decimal number 1 2

2 4
2. Subtract largest power of two less than or 3 8
equal to number
4 16
3. Put a one in the corresponding bit position 5 32
4. Keep subtracting until result is zero 6 64
7 128
5. Append a zero as MS bit; if original was
8 256
negative, flip bits and add +1
9 512

X = 104 ten
10 1,024
104 − 64 = 40 bit 6
40 − 32 = 8 bit 5
8−8=0 bit 3
X = 01101000 two
21
Fractions: Fixed-Point Binary

› We use a "binary point" to separate integer bits from fractional bits,

just like we use the "decimal point" for decimal numbers
› Two's complement arithmetic still works the same

› NOTE: In a computer, the binary point is implicit -- it is not

explicitly represented. We just interpret the values appropriately
22
 Converting Decimal Fraction to Binary (2's C)
1. Multiply fraction by 2
2. Record one's digit of the result, then subtract it from the decimal
number
3. Continue until you get 0.0000... or you have used the desired number
of bits (in which case you have approximated the desired value)

X = 0.421ten 0.421  2 = 0.842 bit − 1 = 0

0.842  2 = 1.684 bit − 2 = 1
0.684  2 = 1.368 bit − 3 = 1
0.368  2 = 0.736 bit − 4 = 0
until 0, or until no more bits...
X = 0.0110 two › (if limited to 4 fractional bits)
23
Arithmetic Operations
Operation: Addition

› As discussed, 2's complement addition is just binary addition

• Assume operands have the same number of bits
• Ignore carry-out

25
Operation: Subtraction

› You can, of course, do subtraction in base-2, but easier to negate

the second operand and add
› Again, assume same number of bits, and ignore carry-out

26
Operation: Sign Extension

› To add/subtract, we need both numbers to have the same

number of bits. What if one number is smaller?
› Padding on the left with zero does not work:

› Instead, replicate the most significant bit (the sign bit):

27
 Overflow

› If the numbers are two big, then we cannot represent the sum using
the same number of bits
› For 2's complement, this can only happen if both numbers are positive
or both numbers are negative

› How to test for overflow:

1. Signs of both operands are the same, AND
2. Sign of the sum is different

28
Hexadecimal Notation: Binary Shorthand

› To avoid writing long (error-prone) binary values, group four bits

together to make a base-16 digit. Use A to F to represent values 10 to 15

binary hexadecimal decimal binary hexadecimal decimal

(base 2) (base 16) (base 10) (base 2) (base 16) (base 10)

0000 0 0 1000 8 8
0001 1 1 1001 9 9
0010 2 2 1010 A 10
0011 3 3 1011 B 11
0100 4 4 1100 C 12
0101 5 5 1101 D 13
0110 6 6 1110 E 14
0111 7 7 1111 F 15

› Example: Hex number 3D6E easier to communicate

than binary 0011110101101110
29
Converting from binary to hex

› Starting from the right, group every four bits together

into a hex digit. Sign-extend as needed.

This is not a new machine representation or data type,

just a more convenient way to write the numbers

30
Very Large or Very Small Numbers: Floating-Point

Large Values: 602300000000000000000000

Small Values: 0.0000000000000000000000000000000006626

› Large values: 6.023  1023 › -- requires 79 bits

› Small values: 6.626  10 −34 › -- requires > 110 bits
• Range requires lots of bits, but only four decimal digits of precision

› Use a different encoding for the equivalent of “scientific

notation”: F × 2E
Very Large or Very Small Numbers: Floating-Point

› Use a different encoding for the equivalent of “scientific

notation”: F × 2E
› Need to represent F (fraction), E (exponent), and sign
› IEEE 754 Floating-Point Standard (32-bits):

32
Floating Point Representation
› The binary number is written in scientific format as
F × 2E
› F is normalized to be between 1 and 2
1≤F<2
› Because F always has the format 1.xxxx, we may
avoid storing 1, and assume its existence.
› 127 is added to E before storing it. Therefore,
negative exponents will be between 1 and 126.
Floating Point Representation
› The sign of the number is given by the left-most bit.
› The smallest valid value of E is 1.
› If E is zero, the existence of 1 before floating point is
not assumed

N = (−1) S 1.fraction  2exponent −127 , 1  exponent  254

N = (−1) S  0.fraction  2−126 , exponent = 0
Floating-point Example

› Single-precision IEEE floating point number:

• Sign is 1 – number is negative

• Exponent field is 01111110 = 126 (decimal)
• Fraction is 0.100000000000… = 0.5 (decimal)

Value = − 1.5  2(126−127 ) = −1.5  2−1 = −0.75.

35
Converting from Decimal to Floating-Point Binary
› How do we represent −6 5 8 as a floating-point binary number?

› Ignoring sign, represent as a binary fraction: 0110.101

› Next, normalize the number, by moving the binary point
to the right of the most significant bit:
0110.101 = 1.10101  22
› Exponent field = 127 + 2 = 129 = 10000001
› Fraction field = everything to the right of the binary point:
101010000....
› Sign field is 1 (because the number is negative).
› Answer:
› 11000000110101000000000000000000

36
Special Values

Infinity
• If exponent field is 11111111, the number represents infinity
• Can be positive or negative (per sign bit)

Subnormal
• If exponent field is 00000000, the number is not normalized. This
means that the implicit 1 is not present before the binary point.
The exponent is −126
−126
• N = (−1)  0. fraction  2 .
S

• Extends the range into very small numbers

• Example: 00000000000001000000000000000000
0.00001two  2−126 = 2−131 Q: smallest number?
37
Special Values
› The IEEE floating point number representation that
strictly enforces the assumption of a leading one bit
cannot store the value zero.
› To handle zero, the IEEE standard makes an
exception:
• when all bits are zero, the implicit assumption is ignored,
and the stored value is taken to be zero.

Example:
00000000000000000000000000000000
Binary Coded Decimal Representation
› Digital computers employ the binary representations
for integers and floating point numbers.
› However, not every decimal fraction can be
represented using binary floating point numbers.
› Example:
• Convert 0.1 (10) to binary => 0.0001100110011….
• 0.1 x 2 = 0.2
• 0.2 x 2 = 0.4
• 0.4 x 2 = 0.8
• 0.8 x 2 = 1.6
• 0.6 x 2 = 1.2
• 0.2 x 2 =0.4
Binary Coded Decimal Representation
› The use of binary fractions has some unintended
consequences, and their use does not suffice for all
computations.
› For example, consider a bank account that stores
Euros and cents.
› Cents are usually represented as hundredths of
Euros.
› 10.83 denotes 10 Euros and 83 cents.
› If binary floating point arithmetic is used for bank
accounts, individual cents are rounded, making the
totals inaccurate.
BCD Codes
› BCD systems represent digits in binary where four
bits are used to represent each digit.
Character Sets
› Bits have no intrinsic meaning; the hardware or
software must determine what each bit represents.
› More than one interpretation can be used; a group of
bits can be created with one interpretation and later
used with another.
› Example:
• character data has both a numeric and symbolic
interpretation.
› Each computer system defines a character set to be a
set of symbols that the computer and I/O devices
agree to use.
Character Sets
› A typical character set contains uppercase and
lowercase letters, digits, and punctuation marks.
› In the 1960s, IBM Corporation chose the Extended
Binary Coded Decimal Interchange Code (EBCDIC)
representation as the character set used on IBM
computers.
› The American National Standards Institute (ANSI)
defined a character representation known as the
American Standard Code for Information
Interchange (ASCII).
› The goal was to make peripheral devices compatible.
ASCII Codes
Summary
› Digital computers use binary number to represent
data.
› Arithmetic operations on binary data can be
performed in sign-magnitude, 1’s complement, or 2’s
complement forms.
› Floating point representation of binary numbers use
sign-exponent-fraction format.
› Floating point representation includes a discrete
subset of real numbers.
› Some decimal numbers cannot be represented using
a limited number of binary digits.
Questions?