Lecture 9

CS 101:
Computer Programming and

Utilization
Puru
with
CS101 TAs and Staff
Course webpage: https://www.cse.iitb.ac.in/~cs101/
Lecture 9: Number representation
Autumn 2019 CS101@CSE IIT Bombay

Number representation and
variables
02/08/19 Autumn 2019 CS101@CSE IIT Bombay 2

why numbers?
• problem solving using computers is about
problem solving on numbers
• examples
– salaries, temperature, length, distance, voltage …
– picture, language, characters, …

Picture Representation and Reconstruction
0 0 0 1 1 1 0 0 0 0
0 0 1 0 0 0 1 1 0 0
0 1 0 0 0 0 0 0 1 0
1 0 0 0 0 0 0 0 1 0
1 0 1 0 0 0 1 0 0 1
1 0 0 0 0 0 0 0 0 1
1 0 0 1 1 1 0 0 1 0
0 1 0 0 0 0 0 0 1 0
0 0 1 0 0 0 1 1 0 0
0 0 0 1 1 1 0 0 0 0
(a) (b) (c)

radix-based number systems
• Number systems with radix r has r symbols (including 0)
• (x)r x is a string of symbols, r is the radix/base
• key idea: position of a symbol determines it's value!
PLACE VALUE
– Multiply from right to left by: r0, r1, r2, r3, ... and then add
• decimal number system (r=10)

• hexa-decimal number system (r=16)
• binary (r=2)

examples
decimal hexa-decimal binary
26
0xf
255
16
0xffff

examples
decimal hexa-decimal binary
26 1a (0x1a) 10110
15 0xf 1111
255 0xff 11111111
16 0x10 00010000
65535 0xffff 1111111111111111

number representation with computers
• all numbers represented in binary form
• data types:
– int (32 bits)
– unsigned int (32 bits)
– float (32 bits)
– double (64 bits)
• binary sequences in each have different formats

two’s complement representation for
integers
• For a n-bit integer (numbers to be stored in n bits)
• If x is positive: (0 <= x <= 2n-1 – 1)

– Binary form of x
• If x is negative ( -2n-1 <= x < 0)

– one’s complement of positive number + 1
– E.g. -9 in 2's complement:
one’s complement (00001001) + 1 = 11110110 + 1 = 11110111
– Binary form of 2n + x
100000000 - 000001001 = 11110111 = 247 decimal

the two’s complement wrap-around
Min=10…0 -1=1…1 0=0…0 Max=01…1
Zero is one position

to the right of center

why two’s complement?
• single representation for zero
• addition is same as subtraction!
• examples:
8 + 4
8-4
4-8
The bool type
• Old C++ used int to store Boolean values

• But ANSI standard C++ offers a type called bool
• bool tval = true, fval = false;
• int ival = int(tval);
• However, old bad habits still allowed
– if (37) { … }
– bool bval = 37;
• Overall value unclear

Bit array manipulation
• Fixed size integers are arrays of bits
• C++ lets you do bitwise Boolean algebra
• a & b (and), a | b (or), a^b (exor), ~b (not)

Bitwise operators
• a & b (and), a | b (or), a^b (exor), ~b (not)
10110110 10110110
10010101 10010101
& ^
10010100 00100011
10110110 00100011
10010101 ~
| 11011100
10110111
Left shift operator
• int c = 5; cout << (c << 2);
00000000,00000000,00000000,00000101
00000000,00000000,00000000,00010100
• Bits lost from the left (msb)

• Zero bits inserted from the right (lsb)
• Result is 20 (= 5 ´ 22)
• Cheap way to multiply by powers of two
Right shift operator
• c >> 2
• Bits discarded to the right (lsb)
• If msb of c was 0, then 0 bits injected from left
(msb)
– 5 >> 2 gives 1
• If msb of c was 1 (c was negative) then 1 bits
injected from left
– -5 >> 2 gives -2 (work it out)
– 0xfffffffb >> 2 gives 0xfffffffe
• Preserves sign of number
examples
• generate powers of 2 using shift operator
• outputs of …
int a = 1;
cout << a;
a = a << 31;
cout << a;
a = a >> 31;
cout << a;

Some applications of bit operations
• Is an int x odd or even?
• Remainder when divided by 8

•
How many one bits in a 32-bit int?

Some applications of bit operations
• Is an int x odd or even?

– int isOdd = (x & 1);
• Remainder when divided by 8
– int remain = (x & 7);
– Faster than x % 8
• How many one bits in a 32-bit int?
• Repeat 32 times:
– numOnes = numOnes + (x & 0x8000000);
– x = x << 1;
In binary this looks like a one
followed by 31 zeros
fractions in binary
• powers on the right side of the point are negative:
8 4 2 1 1/2 1/4 1/8 1/16
• binary 0.1 = 0 + 1 x 2-1 = 0.5 in decimal
• in binary 0.11 = 0x 1 + 1 x 2-1 + 1 x 2-2
= 0.5 + 0.25 = 0.75 in decimal

fractions and precision
• represent decimal number 1.45 using 3 bits and
4 bits

fractions and precision
• represent decimal number 1.45 using 3 bits and
4 bits
• with 3 bits: 1.11 => 1.375

• with 4 bits: 1.111 => 1.4375
• for actual storage as numbers all factions

converted to mantissa + exponent form
• length of mantissa determines precission!
Representing floating point numbers
• Use an analogue of scientific notation:
• significand * 10exponent, e.g. 6.022 * 1022
• Significand (with a bit for sign) and exponent are in binary
• significand * 2exponent
Costs how No of bits for No of bits for

many bits to mantissa exponent
store
float 32 24 8
double 64 52 12
• Actual representation: more complex. “IEEE Floating Point

Standard”
example
• representation of number 65 as a float
• 65 in binary is 1000001
• 65 in significand—exponent form
– 6.5 x 101
– 1000001
– 1.000001 x 2110
– binary point moved 6 places to the right
– 6 is 110 in binary

example, number 65 as a float
• 65 decimal in binary signifand-exponent form
– 1.000001 x 2110
• For float,
– 23 bits for mantissa (+1 bit for sign)
– 7 bits for exponent (+1 bit for sign)
• 65 decimal is stored as a float as follows,
– decimal point is assumed after 2nd bit
– 01000001000000000000000000000110
sign of mantissa sign of exponent

more examples
• 36 as a float
• 3.75 as float

more examples
• 36 as a float
• 36 is 100010 in binary
• 1.00010 x 2101
in binary significand (mantissa)-exponent form
• 01000100000000000000000000000101

more examples
• 3.75 as a float
• integral part is 3: 011 in binary
• fractional part is .75 which is .11 in binary
– 1*(1/2) + 1 * (1/4) = 0.5 + 0.25 = 0.75
• 3.75 decimal is 11.11 binary
• convert to significand-exponent form
– 1.111 x 21
– Store in 32-bit float

more examples
• 3.76
• 0.75
• 3450
– 1.10101111010 x 21011

more examples
• Let us represent the number 3450 = 3.45 x 103
• First: Convert to binary:
• 3450 = 211+ 210+ 28 + 26+ 25+24 +23 + 21
11 10 9 8 7 6 5 4 3 2 1 0
1 1 0 1 0 1 1 1 1 0 1 0
• Thus 3450 in binary = 110101111010

• 3450 in significand-exponent notation: how?
• 1.10101111010 x 21011
− 10 in binary is 2 in decimal
− 1011 in binary is 11 in decimal, we have to move the
"binary point" 11 places to the right
− as float
01101011110100000000000000001011
precision, big numbers
• int, unsigned int, long, long unsigned int
• float, double, long double
• why does double have higher precision than float?

different Needs, different Variable Types
• The keyword long : says, I unsigned int

telephone_number;
need to store bigger or
more precise numbers, so float velocity;
give me more than usual
float mass, acceleration;
space.
• long unsigned int: Likely 64 long unsigned int
crypto_password;
bits will be allocated
• long double: likely 96 bits long double
more_precise_vaule;
will be allocated

the Const Keyword
• const double pi = 3.14;
• The keyword const means : value assigned once cannot be

changed
• Useful in readability of a program

• area = pi * radius * radius;
• reads better than
• area = 3.14 * radius * radius;

The Hardware
(A very high level glimpse)
• How do we store numbers in hardware?
• How is an instruction expressed in hardware?
• How is it executed?
• How do we output the numbers?

Digital Circuits - Operations
• Building blocks of computers
• Circuits have wires that carry current,
and are at a certain electric potential.
• Digital circuits: interpret electrical
potential/voltage as numbers.
• Simplest convention
− Voltage above 5 volt = number 1
− Voltage between 0 and 0.2 volt =
number 0
Copyright © 1999-2000 Michael Stutz stutz@dsl.org − Circuit designed so that voltage will
never be between 0.2 and 5 volt,
An inverter circuit hence no ambiguity.

Digital Circuits - Storage
• Capacitors (like batteries) can store electrical charges
• Charge stored on a capacitor may also denote numbers
− Capacitor has low charge = value 0 stored on it.
− Capacitor has high charge = value 1 stored on it.
− Once charge is stored on a capacitor, it persists. Memory
• Building blocks of DRAMs (Dynamic Random Access Memory)

Representing Numbers
• How to represent numbers using this capability?
• Key idea : Binary number system
– Represent all numbers using only 1's and 0's
– Also called "Bits": "Binary digits"

Representing Numbers
Key idea:
Store each bit of the number on a
separate capacitor
• Example: 25 Decimal = 11001 binary

– Use 5 capacitors
– Store high charge on 1st, 2nd, 5th,
and low charge on 3rd, 4th
• To transmit 25 from one part of the computer
to another
– Use 5 wires and raise the wires to appropriate
voltages at one end.
– Sense the voltages at the other end

Bits, bytes, half-words, words
• Bit = 1 capacitor/wire
• byte = 8 capacitors/wires
• half-word = 16 capacitors/wires
• word = 32 capacitors/wires
• double word = 64 capacitors/wires

Organization of a computer

Memory
• Organized into bytes (groups 0 0 0 0 1 1 0 1 0
of 8 capacitors) 1
• Memories of present day 2
computers contain few 3
Gigabytes, Giga=230 4
• Each byte in the memory is 5 1 0 1 1 1 1 0 1
assigned a distinct number, or 6
an address 7
• In a memory with N bytes, 8
the addresses go from 0 to N- 9
1

Memory Ports
• Memory has 3 ports: address port,
data port, control port.
• Address port consists of log N wires.
(N = number of bytes in memory)
• You can place numbers 0..N-1 on
address port
• Control Port may be just 1 wire.
– Wire = 0: Memory to perform read
operation.
– Wire = 1: Memory to perform write
operation.
• Data port will have w wires, where
w is a multiple of 8. Say w=8m.

Write Operation
• Control Port must be set to 1.
• If A is placed on the address
port, and D on data port, then
D will be stored in the m bytes
starting at byte A.
• (Remember that the data port

had 8m wires, and so m bytes
are available on the data port)

Write Operation
10010
11101000
"Write the number

11101000 (232) into
0 1 the location number
10010 (18)
Read Operation
• Control Port must be

set to 0.
• If A is placed on the
address port, the m
bytes starting at byte A
will appear on the data
port
• (Data port has 8m
wires, and so m bytes
will fit on the data port)

Read Operation
10010
11101000
"Read from the

location number
1 0 10010 (18)

Arithmetic Unit
• Ports: Input1, Input2, Output, Control

• Typically Input1, Input2, Output will consist of w wires,
w = size of memory data port
• Control = several wires. Number appearing on the control
wires will say what operation should be performed
• 1 cycle after values are placed on Control, the Output will take
the commanded value: sum, product, …

Peripherals: keyboard, screen, disk…
• Also have control port
and data port like data
organization.
• Depending upon value control
placed on control port, data
the peripheral decides
what to do with the control
value on the data port, data
or itself places values on
the data port. control

Control Unit
• Tells other parts of the computer what to
do.
– Sends numbers on control wires of each unit
• The control unit decides what to tell other

units by reading a “machine language
program” from the memory of the
computer.

Machine language program
OPCODE OPERAND1 OPERAND2 OPERAND3
• Program = Sequence of instructions

• Instruction = sequence of numbers
– First number is OPERATION CODE (OPCODE). This is the
code that tells the Control Unit what OPERATION should
do.
– Subsequent numbers are OPERANDS. These are
"arguments" to the operation
• i386
• x86_64
• PowerPC
Example
57 100 200 300
• operation code 57 might mean:

– Interpret the 3 numbers following 57 as addresses.
– Read the words at the first two addresses and send them
to the Arithmetic unit.
– Command the arithmetic unit to perform multiplication by
sending appropriate number on its control wires.
– Store the result from the arithmetic unit into the word at
the third address

Machine language progam
• Program = Sequence of instructions
• Instruction = sequence of numbers, first of which is an
operation code which denotes what operation to
perform
• Example: operation code 57 might mean:
– Interpret the 3 numbers following 57 as addresses.
– Read the words at the first two addresses and send them to the Arithmetic unit.
– Command the arithmetic unit to perform multiplication by sending appropriate number on its control wires.
– Store the result from the arithmetic unit into the word at the third address
• The sequence 57, 100, 200, 300 is an instruction that

would cause the product of the numbers stored in
addresses 100, 200 to be stored in the address 300.
• The operation codes are defined by the computer
designer.
– She will assign a different code for each operation that she would like the computer to perform.
– Example: 58 might mean the same thing as above, except that the numbers would be added.

Control unit operation
• Control unit must be told where the machine
language program is in memory
• The control unit then fetches the instructions

constituting the program, interprets the codes,
and performs the required operation
• After one instruction is fetched and executed, it

fetches the next instruction and repeats the
process

Putting it together
Arthimetic 1. Control unit is told
Unit
the address of the
0
1 instruction to fetch
.. (e.g. instruction is at
10 57 100 200 300 location 10)
...
10
100 289 2. A read operation
.... is performed on
memory location 10
200 345 Bus
(network)
...
57 100 200 300
3. instruction
300 57 100 200 300
Control Unit
is now "loaded" into
Memory the control unit
1 0
Read Write

Putting it together
Arthimetic 1. Now control unit
Unit
performs a read
0
1 operation of address
.. 100
10 57 100 200 300
...
100
100 289
2. The number 289
.... is sent to input 1 of
arithmetic unit
200 345 Bus
289 (network)
...
300
Control Unit
Memory
1 0
Read Write

Putting it together
Unit
performs a read
0
1 operation of address
.. 200
10 57 100 200 300
...
200 2. The number 345
100 289
is sent to input 2 of
.... arithmetic unit
200 345 Bus
345 (network)
...
300
Control Unit
Memory
1 0
Read Write

Putting it together
Unit
instructs the
0 99705
1 arithmetic unit to
.. x perform a multiply
10 57 100 200 300 operation (the
...
"control" wires are
100 289
set to the operation
.... code of "multiply")
200 345 Bus
(network)
2. Control unit
... stores the product
300 99705 (289 x 345 = 99705)
x Control Unit temporarily inside its
Memory own unit
0 0
Read Write

Putting it together
Unit
performs a write
0
1 operation on
.. address 300
10 57 100 200 300
...
300 2. The number
100 289
99705 is sent on the
.... data port of memory
and 300 is sent on
200 345 Bus
99705
(network)
the address port of
... the memory
300 99705
Control Unit
Memory
0 1
Instruction
execution is
Read Write COMPLETE

Lecture 9

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture 9

Uploaded by

Copyright:

Available Formats

CS 101:

Computer Programming and

Course webpage: https://www.cse.iitb.ac.in/~cs101/

Lecture 9: Number representation

Autumn 2019 CS101@CSE IIT Bombay

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 2

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 3

(a) (b) (c)

Autumn 2019 CS101@CSE IIT Bombay

• decimal number system (r=10)

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 5

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 6

255 0xff 11111111

65535 0xffff 1111111111111111

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 7

• binary sequences in each have different formats

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 8

• If x is positive: (0 <= x <= 2n-1 – 1)

• If x is negative ( -2n-1 <= x < 0)

Autumn 2019 CS101@CSE IIT Bombay

Min=10…0 -1=1…1 0=0…0 Max=01…1

Zero is one position

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 10

• Old C++ used int to store Boolean values

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 12

• C++ lets you do bitwise Boolean algebra

• a & b (and), a | b (or), a^b (exor), ~b (not)

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 13

• int c = 5; cout << (c << 2);

• Bits lost from the left (msb)

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 17

• Is an int x odd or even?

• Remainder when divided by 8

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 18

• Is an int x odd or even?

• powers on the right side of the point are negative:

8 4 2 1 1/2 1/4 1/8 1/16

• binary 0.1 = 0 + 1 x 2-1 = 0.5 in decimal

• in binary 0.11 = 0x 1 + 1 x 2-1 + 1 x 2-2

= 0.5 + 0.25 = 0.75 in decimal

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 20

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 21

• with 3 bits: 1.11 => 1.375

• for actual storage as numbers all factions

Costs how No of bits for No of bits for

• Actual representation: more complex. “IEEE Floating Point

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 25

sign of mantissa sign of exponent

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 27

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 28

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 29

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 30

• Thus 3450 in binary = 110101111010

• float, double, long double

• why does double have higher precision than float?

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 32

• The keyword long : says, I unsigned int

02/08/19 Autumn 2019 CS101@CSE IIT Bombay 33

• const double pi = 3.14;

• The keyword const means : value assigned once cannot be

• Useful in readability of a program

• reads better than