Data Representation

CS2052 Computer Architecture

Computer Science & Engineering
University of Moratuwa

Dilum Bandara
[email protected]
Outline
 Representing numbers
 Unsigned
 Signed
 Floating point
 Representing characters & symbols
 ASCII
 Unicode

2
Data Representation in Computers
 Data are stored in Registers
 Registers are limited in number & size

n-1 n-2 … ... ... 2 1 0

n-bits

 With a n-bit register

 Min value 0
 Max value 2n-1

3
Data Representation
Qualitative
• Represents quality or
characteristics
• Not proportional to a
value
Data • Name, NIC no, index no,
Representation Address
Quantitative
• Quantifiable
• Proportional to value α
• No of students, marks for
CS2052, GPA 4
Data Representation (Cont.)
Signed

Integers
Unsigned

Quantitative

Signed
Data
Non-
integers
Unsigned
Qualitative

5
Number Systems
 Decimal number system
 0, 1, 2, 3, 4, 5, 6, 7, 8, 9
 Binary number system
 0, 1
 Octal number system
 0, 1, 2, 3, 4, 5, 6, 7
 Hexadecimal number system
 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, A, B, C, D, E, F

6
Quantitative Numbers
 Integers
 Unsigned 20
 Signed +20, -20
 Non-integers
 Floating point numbers - 10.25, 3.33333…, 1/8 = 0.125

7
Signed Integers
 We need a way to represent negative values
 3 representations
 Sign & Magnitude representation (S&M)
 Complement method
 Bias notation or Excess notation

8
1. Sign & Magnitude Representation

sign n-1 n-2 ... ... 2 1 0

Magnitude (n-bits)

 n-bit unsigned magnitude & sign bit (S)

 If S
 0 – Integer is positive or zero
 1 – Integer is negative or zero
 Range –(2n-1) to +(2n-1)

9
Example – Sign & Magnitude
 If 8-bit register is used what are min & max
numbers?
 What are 0000 0000 and 1000 0000 in decimal?
 Representation of zero is not unique

10
Sign & Magnitude (Cont.)
 Advantages
 Sign reversal
 Finding absolute value |a|
 Flip sign bit
 Disadvantage
 Adding a negative of a number is not the same as
subtraction
 e.g., add 2 and -3
 Need different operations

 Zero is not unique

11
2. Complement Method
 Base = Radix
 Radix r system means r number of symbols
 e.g., binary numbers have symbols 0, 1
 2 types
 r’s complement
 (r – 1)’s complement
 Where r is radix (base) of number system
 Examples
 Decimal 9’s & 10’s complement
 Binary 1’s & 2’s complement

12
Complement Method – Definition
 Given a number m in base/radix r & having n
digits
 (r – 1)’s complement of m is
(rn – 1) – m
 r ’s complement of n is
(rn – 1) – m + 1 = rn – m

13
Example – Complement Method
 If m = 5982 & n = 4 digits
 9’s complement is
9999 - maximum representable no
5982–
4017
 10’s complement
9999 or 10000
5982– 5982–
4017 4018
1+
4018
14
Example – Complement Method
 If m = 382 & n = 3
 -n = -382 =
999 999
382– 382-
617 617
1+
618
 -382 = 617 or 618
 Depending on which complement we use
 These are called complementary pair

15
1’s complement
 Calculated by
 (2n – 1) – m
 If m = 0101
 1’s complement of m on a 4-bit system
1111
0101–
1010
 This represents -5 in 1’s complement

16
Finding 1’s Complement – Short Cut
 Invert each bit of m
 Example
 m= 00101011
 1’s complement of m = 11010100

 m= 00000000 = 0
 1’s complement of m = 1 1 1 1 1 1 1 1 = -0

 Values range from -127 to +127

17
Addition with 1’s Complement
 If results has a carry add it to LSB (Least
Significant Bit)
 Example
 Add 6 and -3 on a 3-bit system
 6 = 110
 -3 = 100+
= 1010
1+
011

18
2’s Complement
 Doesn’t require end-around carry operation as in
1’s complement
 2’s complement is formed by
 Finding 1’s complement
 Add 1 to LSB
 New range is from -128 to +127
 -128 because of +1 to negative value

19
Example – 2’s Complement
 Find 2’s complement of 0101011
 m =0101011
 =1010100
________ 1+
 2’s =1010101

 Short-cut
1. Search for the 1st bit with value 1 starting from LSB
2. Inver all bits after 1st one

20
Example – 2’s Complement (Cont.)
 Add 10 and -5 on a 4-bit system
5 = 0101
-5 = 1011
10 = 1010
-5 = 1 0 1 1+
10101  0101
Discard

21
1’s vs. 2’s Complement
 1’s complement has 2 zeros (+0, -0)
 Value range is less than 2’s complement

 2’s complement only has a single zero

 Value range is unequal

 No need of a separate subtract circuit

 Doing a NOT operation is much more cost effective
interms of circuit design
 However, multiplication & division is slow
22
 S&M, 1’s, & 2’s

23
Source: www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html
4-bit, 2’s Complement Adder
Subtractor

Source: www.instructables.com/id/How-to-Build-an-8-Bit-Computer/step7/The-
ALU/ 24
Detecting Negative Numbers &
Overflow
 Check for MSB
 To find magnitude
 1’s complement
 Flip all bits
 2’s complement
 Flip all bits + 1
 Rules to detect overflows in 2’s complement
 If sum of 2 positive numbers yields a negative result,
sum has overflowed
 If sum of 2 negative numbers yields a positive result,
sum has overflowed
 Else, no overflow 25
3. Bias Notation
 Number range to be represented is given a
positive bias so that smallest unsigned value is
equal to -B
 If bias is B
 Value B in number range becomes zero
n

b 2
i 0
i
i
 B

 If bias is 127 for 8-bit number range is

 -127 (0 - 127) to 128 (255 - 127)
26
Bias Notation (Cont.)
Bit pattern Unsigned Biased-127
0000 0000 +0 -127
0000 0001 +1 -126
0000 0010 +2 -125
… … …
0111 1110 +126 -1
0111 1111 +127 +0
1000 0000 +128 +1
… … …
1111 1111 +255 +128 27
Floating Point Numbers
 We needed to represent fractional values &
values beyond 2n – 1
 +3207.23 = 3.20723x103
 -0.000321 = -3.21x10-4 Exponent

Radix
Sign (base)
Mantissa

28
Floating Point Numbers (Cont.)
Sign
s e
N ( 1) m.2 Exponent

Radix

Mantissa

sign en-1 … ... e0 mn-1 mn-2 … … m1 m0

Exponent Mantissa

29
IEEE Floating Point Standard (FPS)
 2 standards
1. Single precision
 32-bits 31 30 23 22 0
 23-bit mantissa S Exponent Mantissa (bits 0-22)
 8-bit exponent

 1-bit sign

2. Double precision
 64-bits
 52-bit mantissa
 11-bit exponent 63 62 52 51 0
 1-bit sign
S Exponent Mantissa (bits 0-51)

30
IEEE Floating Point Standard (Cont.)
s31 e30 … ... e23 m22 mn-2 … … m1 m0

Exponent Mantissa

s E  127
N ( 1) [1.m] 2
 E – 127 = e
 E = e + 127
 What is stored is E

31
Single Precision
 23-bit mantissa
 m ranges from 1 to (2 – 2-23 )
 8-bit exponent
 E ranges from 1 to 254
 0 & 255 reserved to represent -∞, +∞, NaN
 e ranges from -126 to +127
 Maximum representable number
 2 x 2127 = 2128

32
Review – Binary Fractions
 What is 101.11012 in decimal?
 22 x 1 + 21 x 0 + 20 x 1 + 2-1 x 1 + 2-2 x 1 + 2-3 x 0 + 2-4 x 1
 = 4 x 1 + 2 x 0 + 1 x 1 + 0.5 x 1 + 0.25 x 1 + 0.125 x 0 + 0.0625 x 1
 = 4 + 1 + 0.5 + 0.25 + 0.0625
 = 5.8125
 What is 5.812510 in binary?
 Integer – 5 = 101
 Fraction – Keep multiplying fraction until answer is 1
 0.8125 x 2 = 1.625
 0.625 x 2 = 1.25

 0.25 x 2 = 0.5
 0.5 x 2 = 1.0
 0.8125 = .1101 33
10 2
Example – Single Precision
 IEEE Single Precision Representation of 17.15 10
 Find 17.1510 in binary
17.1510 = 10001.00100112
= 1.00010010011 x 24
E = e + 127
E = 4 + 127 = 131

0 1000 0011 0001 0010 0110 0000 ……..

Exponent Mantissa

34
Character Representation
 Belong to category of qualitative data
 Represent quality or characteristics
 Includes
 Letters a-z, A-Z
 Digits 0-9
 Symbols !, @ , *, /, &, #, $
 Control characters <CR>, <BEL>, <ESC>, <LF>

35
Character Representation (Cont.)
 With a single byte (8-bits) 256 characters can be
represented
 Standards
 ASCII – American Standard Code for Information
Interchange
 EBCDIC – Extended Binary-Coded Decimal
Interchange Code
 Unicode

36
ASCII
 De facto world-wide standard
 Used to represent
 Upper & lower-case Latin letters
 Numbers
 Punctuations
 Control characters
 There are 128 standard ASCII codes
 Can be represented by a 7 digit binary number
 000 0000 through 111 1111
 Plus parity bit

37
ASCII Table
ASCII ASCII Hex Symbol ASCII Hex Symbol
Hex Symbol
0 0 NUL 32 20 (space) 48 30 0
1 1 SOH 33 21 ! 49 31 1
2 2 STX 34 22 " 50 32 2
3 3 ETX 35 23 # 51 33 3
4 4 EOT 36 24 $ 52 34 4
5 5 ENQ 37 25 % 53 35 5
6 6 ACK 38 26 & 54 36 6
7 7 BEL 39 27 ' 55 37 7
8 8 BS 40 28 ( 56 38 8
9 9 TAB 41 29 ) 57 39 9
10 A LF 42 2A * 58 3A :
11 B VT 43 2B + 59 3B ;
12 C FF 44 2C , 60 3C <
13 D CR 45 2D - 61 3D =
14 E SO 46 2E . 62 3E >
15 F SI 47 2F / 63 3F ?

38
ASCII Table (Cont.)
ASCII Hex Symbol ASCII Hex Symbol ASCII Hex Symbol
64 40 @ 80 50 P 96 60 `
65 41 A 81 51 Q 97 61 a
66 42 B 82 52 R 98 62 b
67 43 C 83 53 S 99 63 c
68 44 D 84 54 T 100 64 d
69 45 E 85 55 U 101 65 e
70 46 F 86 56 V 102 66 f
71 47 G 87 57 W 103 67 g
72 48 H 88 58 X 104 68 h
73 49 I 89 59 Y 105 69 i
74 4A J 90 5A Z 106 6A j
75 4B K 91 5B [ 107 6B k
76 4C L 92 5C \ 108 6C l
77 4D M 93 5D ] 109 6D m
78 4E N 94 5E ^ 110 6E n
79 4F O 95 5F _ 111 6F o

39
ASCII – Things to Note
 ASCII codes for digits aren’t equal to numeric
value
 Uppercase & lowercase alphabetic codes differ
by 0x20
 Shift key clears bit 5
 0x20 = 32 = 0010 0000
 Example:
 A = 65 = 0100 0001
 a = 97 = 0110 0001

 Most languages need more than 128 characters

40
Unicode (www.unicode.org)
 Designed to overcome limitation of number of
characters
 Assigns unique character codes to characters in
a wide range of languages
 A 16-bit character set
 UCS-2
 UCS-4 is 32-bit
 65,536 (216) distinct Unicode characters
Unicode provides a unique number for every character,
no matter what the platform,
no matter what the program,
no matter what the language 41
 Unicode (Cont.)

Source: www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

42
Unicode – Sinhala & Tamil

43