CSC10009 – Computer Systems

Chapter 2
Data Representation

Instructor: Vu Thi My Hang, Ph.D.

TA: Le Quoc Hoa, M.Sc.

4.0
fit@hcmus
 4.0
fit@hcmus
Outline

▪ Integer Representation
▪ Integer Arithmetic
▪ Floating Point
▪ Character

2
 4.0
fit@hcmus
Outline

▪ Integer Representation
▪ Integer Arithmetic
▪ Floating Point
▪ Character

3
 4.0
fit@hcmus
(Positional) Number Systems
▪ General formular for unsigned integer with n digits in q-base system
xn-1: MSB x0: LSB
(Most Significant Bit) (Less Significant Bit)

Value Exponential Form

Binary number system (base-2)

Octal number system (base-8)
Decimal number system (base-10)
Hexadecimal number system (base-16)

4
 4.0
fit@hcmus
(Positional) Number Systems (2)
▪ Decimal number

▪ Binary number

▪ Hexadecimal number

5
 4.0
fit@hcmus
(Positional) Number Systems (3)

6
 4.0
fit@hcmus
Converting Binary ó Decimal
▪ Decimal è Binary

▪ Binary è Decimal

7
 4.0
fit@hcmus
Converting Hexadecimal ó Decimal
▪ Decimal è Hexadecimal
Quotient Remainder
123/16 = 7 12 (B)

7/16 =0 7 è 12310 = 7B16

▪ Hexadecimal è Decimal

8
 4.0
fit@hcmus
Converting Hexadecimal ó Binary
▪ Hexadecimal ó Binary

9
 4.0
fit@hcmus
Integer Representation
▪ Unsigned integer A of n-bit:

▪ Signed integer A of n-bit in Sign/Magnitude representation:

an-1(MSB): Sign
0: positive number
1: negative number
an-2 … a0: Magnitude

-2n-1+1 2n-1-1 2n-1

2 zeros: +0/-0

10
 4.0
fit@hcmus
Integer Representation (2)
▪ Signed integer A of n-bit in 2’s Complement representation:

if A >= 0

an-1(MSB): 0 à Positive number

if A < 0 an-1(MSB): 1 à Negative number

-2n-1 2n-1-1 2n-1

11
 4.0
fit@hcmus
Integer Representation (3)
▪ Signed integer A of n-bit in 2’s Complement representation:

12
 4.0
fit@hcmus
Integer Representation (4)

13
 4.0
fit@hcmus
Integer Representation (5)
▪ Signed integer A of n-bit in Bias (Excess-k) representation:
Ø Choose K (a positive integer) to allows operations on the biased
numbers to be the same as for unsigned integers but represents both
positive and negative values. K is usually 2n-1 - 1 or 2n-1.
Ø Positive number: N > K with the absolute value (N – K).
Ø Negative number: N < K with the absolute value (K - N).

▪ E.g.: n = 8; K = 2n-1 -1 = 127

100000012 with unsigned value is 129 (greater than K)
è Positive number
è Absolute value: 129 – 127 = 210

14
 4.0
fit@hcmus
Integer Representation (6)

15
 4.0
fit@hcmus
Sign Extension
▪ Positive number 16-bit to 32-bit

▪ Negative number 16-bit to 32-bit

16
 4.0
fit@hcmus
Outline

▪ Integer Representation
▪ Integer Arithmetic
▪ Floating Point
▪ Character

18
 4.0
fit@hcmus
Logic Operation

19
 4.0
fit@hcmus
Logic Operation (2)

20
 4.0
fit@hcmus
Shift & Rotate

21
 4.0
fit@hcmus
Shift & Rotate (2)

or

22
 4.0
fit@hcmus
Shift & Rotate (3)

Clear a bit (AND with 0 is always 0)

Set a bit (OR with 1 is always 1)
Flip a bit (XOR with 1 = flip that bit)

: Get the value at bit i of x

: Set bit i of x to 1
: Set bit i of x to 0
: Flip bit i of x

23
 4.0
fit@hcmus
Negation
▪ Negation operation ó Two complement operation

24
 4.0
fit@hcmus
Addition
▪ General Rule

▪ Example

25
 4.0
fit@hcmus
Addition (2)

OVERFLOW RULE: If two numbers are added, and they are both positive or both
negative, then overflow occurs if and only if the result has the opposite sign.

26
 4.0
fit@hcmus
Subtraction
▪ General Rule
A – B = A + (-B) = A + (2’s Complement of B)

▪ Example

27
 4.0
fit@hcmus
Subtraction (2)

OVERFLOW RULE: If two numbers are added, and they are both
positive or both negative, then overflow occurs if and only if the
28
result has the opposite sign.
 4.0
fit@hcmus
Multiplication
▪ General Rule

▪ Example

29
 4.0
Multiplication (2)
fit@hcmus (Unsigned Number)

A – B = A + (-B) = A + (2’s Complement of B)

30
 4.0
Multiplication (3)
fit@hcmus (Signed Number: Booth’s Algorithm)

31
 4.0
Division
fit@hcmus (Unsigned Number) Q=7, M = 3

32
 4.0
Division (2)
fit@hcmus (Signed Number)

T = Q/M; R = Q%M
Q = 7, M = 3 è T = 2, R = 1
Q = 7, M = -3 è T = -2, R = 1
Q = -7, M = 3 è T = -2, R = -1
Q = -7, M = -3 è T = 2, R = -1

Q M T R
>0 >0 >0 >0
>0 <0 <0 >0
<0 >0 <0 <0
<0 <0 >0 <0

33
 4.0
fit@hcmus
Outline

▪ Integer Representation
▪ Integer Arithmetic
▪ Floating Point
▪ Character

34
 4.0
fit@hcmus
Fixed-point Numbers
▪ Representation

Integer part Fraction part

▪ Convert decimal è binary

35
 4.0
fit@hcmus
Fixed-point Numbers (2)

Integer part Fraction part

123.37510 = 0111 1011.0110 00002

13.62510 = 0000 1101.1012

36
 4.0
fit@hcmus
Normalized Float Number

±1.F * 2E

Significand (Fraction) Exponent

37
 4.0
fit@hcmus
Normalized Float Number (2)

±1.F * 2E

Significand (Fraction) Exponent

38
 4.0
fit@hcmus
Overflow vs. Underflow

±1.F * 2E

Significand (Fraction) Exponent

§ Overflow: A positive exponent becomes too large to fit in the exponent field.
§ Underflow: A negative exponent becomes too large to fit in the exponent field.

39
 4.0
fit@hcmus
IEEE-754 Standard

Format Sign Exponent Significand

32-bit 1 bit 8 bits 23 bits
64-bit 1 bit 11 bits 52 bits
128-bit 1 bit 15 bits 112 bits

▪ Sign: 1 (negative number), 0 (positive number)

▪ Exponent (Biased representation): Bias K = 2n-1 - 1
o Single Precision (32-bit) K = 127 à Exponent = E + 127
o Double Precision (64-bit) K = 1023 à Exponent = E + 1023
▪ Significand
o Represent the mantissa (fractional part after the binary point)
o Normalized form: 1.xxxxxx

40
 4.0
fit@hcmus
IEEE-754 Standard (2)
Single Precision (32-bit)
X = -5.25 1 1000 0001 0101 0000 0000 0000 0000 000

Step 1: Convert X = -5.2510 to Binary

X = −5.2510 = −101.012
Step 2: Normalize to the form ±1.F × 2ᴱ
X = −5.2510 = −101.012 = −1.0101×22
Step 3: Floating Point Representation
• Negative number → Sign bit = 1
• Exponent (with bias K = 127) E = 2 + 127 = 12910 = 1000 00012
• Significand (fraction part) 0101 0000 0000 0000 0000 000

41
 4.0
fit@hcmus
IEEE-754 Standard (3)
Single Precision (32-bit)
X = 12.625 0 1000 0010 1001 0100 0000 0000 0000 000

Step 1: Convert X = 12.62510 to Binary

X = 12.62510 = 1100.1012
Step 2: Normalize to the form ±1.F × 2ᴱ
X = 12.62510 = 1100.1012 = +1.100101×23
Step 3: Floating Point Representation
• Positive number → Sign bit = 0
• Exponent (with bias K = 127) E = 3 + 127 = 13010 = 1000 00102
• Significand (fraction part) 1001 0100 0000 0000 0000 000

42
 4.0
fit@hcmus
IEEE-754 Standard (4)
Single Precision (32-bit)
X = -3050 1 1000 1010 0111 1101 0100 0000 0000 000

Step 1: Convert X = -5.2510 to Binary

X = −305010 = −1011 1110 10102
Step 2: Normalize to the form ±1.F × 2ᴱ
X = −305010 = −1011 1110 10102 = −1.01111101010×211
Step 3: Floating Point Representation
• Negative number → Sign bit = 1
• Exponent (with bias K = 127) E = 11 + 127 = 13810 = 1000 10102
• Significand (fraction part) 0111 1101 0100 0000 0000 000

43
 4.0
fit@hcmus
IEEE-754 Standard (5)

44
 4.0
fit@hcmus
IEEE-754 Standard (6)
Special numbers
Sign Exponent Significand (Fraction)
Zero 0 or 1 00…0 00…0
Denormalized 0 or 1 00…0 !=0
Plus Infinity (+∞) 0 11…1 00…0
Minus Infinity (-∞) 1 11…1 00…0
NaN 0 or 1 11…1 !=0

45
 4.0
fit@hcmus
Outline

▪ Integer Representation
▪ Integer Arithmetic
▪ Floating Point
▪ Character

46
 4.0
fit@hcmus
Character: ASCII
American Standard Code for Information Interchange

47
 4.0
fit@hcmus
Reference
BOOKS
1. Computer Systems: A Programmers Perspective, Prentice Hall, 2016, Pearson.

2. Computer Organization and Design: The Hardware/Software Interface (5th ed),

D. A. Patterson and J. L. Hennessy, 2014, Morgan Kaufmann.

3. Computer Organization: A Quantitative Approach, Patterson and J. L.

Hennessy, 2017, Morgan Kaufmann.

4. Digital Design and Computer Architecture (2th ed), D.M Harris and S. L. Harris,
2013, Morgan Kaufmann.

48
 4.0
fit@hcmus
4.0

Thank you for your

attention!

49