Floating point

OCR arithmetic
A Level Unit 6
Data types
Computer
Science
H446 – Paper 1

4
Objectives
• Represent positive and negative numbers with a
fractional part in floating point form
• Normalise un-normalised floating point numbers with
positive or negative mantissas
• Add and subtract floating point numbers
Floating point arithmetic
Unit 6 Data Types

Fixed point
• Fixed point binary numbers have a pre-determined
number of bits before and after the point
• This makes fixed point numbers easier to process
but they cannot represent the range or accuracy of
numbers that may be required
Floating point arithmetic
Unit 6 Data Types

Fixed point binary numbers

• We could hold an 8-bit number in fixed point format
like this:

• If the first bit is a sign bit representing -32, what is

the largest positive number that can be held?
• If we want to hold larger numbers, the point must be
set further to the right
• What effect does this have on accuracy?
Floating point arithmetic
Unit 6 Data Types

Fixed point binary numbers

• Suppose you want to convert the decimal number
4.6 to binary using this representation

• The closest you can get is 00100.100, i.e. 4.5

• There is a rounding error of 0.1.
• If you need more accuracy, the binary point would
need to be moved left
Floating point arithmetic
Unit 6 Data Types

Floating point numbers

• When decimal numbers become very large, they are
held in the format m x 10n where m is known as the
mantissa, and n is the exponent
• The number 75,000 can be represented as 0.75 x 105
• How would you represent 458.675?
• How could you represent 0.005 in this format?
Floating point arithmetic
Unit 6 Data Types

Floating point numbers

• When decimal numbers become very large, they are
held in the format m x 10n where m is known as the
mantissa, and n is the exponent
• The number 75,000 can be represented as 0.75 x 105
• How would you represent 458.675? This could be
represented as 0.458675 x 103 or 4.58675 x 102, for
example
• How could you represent 0.005 in this format? This
could be represented as 0.5 x 10-2 or 5 x 10-3
Floating point arithmetic
Unit 6 Data Types

Floating point numbers

• Binary numbers are typically held in 32 or 64 bits
• In our examples we will use just 12 bits, with the first
bit being a sign bit

Sign
bit Mantissa Exponent

0 1 0 1 1 0 1 0 0 0 1 1
• The number above represents the number
0.1011010 x 23 since the exponent is 3 in
decimal
101.1010
• This is 101.1010 or 4 + 1 +0.5 + 0.125
Floating point arithmetic
Unit 6 Data Types

Floating point binary to denary

e.g. convert 0.1101100 0100 to denary, assuming an
8-bit mantissa and 4 bit exponent
• Write down the mantissa, 0.1101100
• Translate the exponent from binary to decimal; 0100 = 4
• The binary point of the mantissa has to be moved 4 places right
to multiply it by 24
• Binary number is 01101.100. Translate using the table below:

• Answer 13.5
Floating point arithmetic
Unit 6 Data Types

Negative binary to denary

• A negative floating point number will start with a 1
(the sign bit)
• The number 1.0010100 0011 has a negative
mantissa and a positive exponent
• Find the two’s complement of the mantissa
• Translate the exponent to denary, giving 3
• The binary point has to be moved 3 places right to multiply it by
23
• What is the binary number and its decimal equivalent?
Floating point arithmetic
Unit 6 Data Types

Negative binary to denary

• Here’s the answer:
• The number 1.0010100 0011 has a negative
mantissa and a positive exponent
• The mantissa is 1.0010100
• The two’s complement of 1.0010100 is 0.1101100
• The exponent is 3
• The mantissa, with the binary point moved 3 places right, is
0110.1100
• The number is -6.75
Floating point arithmetic
Unit 6 Data Types

Negative mantissa
• Try this one:

1 . 0101011 0101
• Find the two’s complement of the mantissa (0.1010101)
• Translate the exponent to decimal giving 5
• Move the binary point 5 places right
• Do you get an answer of -21.25?
Floating point arithmetic
Unit 6 Data Types

Negative mantissa
• Try this one:

1 . 0101011 0101
• Find the two’s complement of the mantissa (0.1010101)
• Translate the exponent to decimal giving 5
• Move the binary point 5 places right 10101.01
• Do you get an answer of -21.25?
Floating point arithmetic
Unit 6 Data Types

Negative exponent
• The leftmost bit of the exponent is a sign bit, just as
it is in the mantissa
• An exponent of 1000 in a 4-bit exponent is -8
• You can calculate the denary value of a negative
exponent, for example 1011, as -8 + 3 = -5
• Or, you can use the twos complement method which
will give you the same result
• What is the denary value of negative exponent 1111?
Floating point arithmetic
Unit 6 Data Types

Negative exponent
• If the exponent is negative, the binary point must be
moved left instead of right
e.g. convert 0.1100000 1110 to denary,
assuming an 8-bit mantissa and 4 bit exponent
• Write down the mantissa, 0.1100000
• Translate the exponent from binary to decimal; 1110 = -2
• The mantissa has to be moved 2 places left to divide it by 22
• Binary number is 0.0011000 = 0.125 + 0.0625 = 0.1875
Floating point arithmetic
Unit 6 Data Types

Worksheet 4
• Now try the questions in Task 1 of the worksheet
Floating point arithmetic
Unit 6 Data Types

Normalisation
• Normalisation is the process of moving the binary
point of a floating point number to provide the
maximum level of precision for a given number of
bits
• This is achieved by ensuring that the first digit after
the binary point is a significant digit
Floating point arithmetic
Unit 6 Data Types

Normalisation
• Here are some examples in denary:
0.1234567 x 107 = 1,234,567
0.0123456 x 108 = 1,234,560
0.0012345 x 109 = 1,234,500
• The first representation with a significant (non-zero)
digit after the decimal point has the maximum
precision using the same number of digits
Floating point arithmetic
Unit 6 Data Types

Normalisation
• In normalised floating point form:
A positive number has a sign bit of 0
and the next digit is always 1
• This means that the mantissa of a positive number
in normalised form always lies between ½ and 1
Floating point arithmetic
Unit 6 Data Types

Example
• Normalise the binary number 0.0011010 0100
• The binary point needs to move 2 places to the right so that
there is a 1 following the binary point
• Pad with 0s at the righthand end, giving 0.1101000
• To compensate for making the mantissa larger, we must
make the exponent smaller
• Subtract 2 from the exponent making it 0010
• The normalised number is 0.1101000 0010

Try this one:

•Normalise the binary number 0.0000111 0111
Floating point arithmetic
Unit 6 Data Types

Answer to example
• Normalising the number 0.0000111 0111
means moving the binary point 4 places right, and
subtracting 4 from the exponent
• This gives 0.1110000 0011
Floating point arithmetic
Unit 6 Data Types

Normalised negative numbers

• An unnormalised negative number will have a sign
bit of 1 and one or more 1s after the binary point
• e.g. 1.1100011 0011
• A normalised negative number has a sign bit of 1
followed by a zero after the binary point
Floating point arithmetic
Unit 6 Data Types

Normalising a negative number

Example: normalise the number 1.1100011 0011
• Move the binary point 2 places right and subtract 2
from the exponent
111.00011 0001
• This is the same as 1.0001100 0001
• Note that leading 1s in a negative number do not
change the value of the number (like leading 0s in a
positive number)
Floating point arithmetic
Unit 6 Data Types

From denary to normalised

binary floating point
Eg: Convert 88 to normalised floating point binary
• First convert the number to fixed point binary
• In fixed point binary, the number is 01011000
• Move the binary point 7 places left
• Set the exponent equal to 7

• The number is 0.1011000 0111

Exercise: Now try converting 17.25 to normalised
floating point binary
Floating point arithmetic
Unit 6 Data Types

Answer to exercise
• Convert 17.25 to normalised floating point binary
• Convert to fixed point binary 10001.010
• Move binary point 5 places left
• Set exponent to 5

• Binary number is 0.1000101 0101

• Carry out the opposite conversion to check the result
Floating point arithmetic
Unit 6 Data Types

Converting a negative denary

number
• If the number is negative, calculate the two’s
complement before normalising
• e.g. calculate the binary equivalent of -17.75
• The number is (-) 010001.11
• One’s complement 101110.00
• Two’s complement 101110.01
• Move the point 5 places left
• Set the exponent equal to 5

• The number is 1.0111001 0101

Floating point arithmetic
Unit 6 Data Types

Worksheet 4
• Now try Task 2 on the worksheet
Floating point arithmetic
Unit 6 Data Types

Adding floating point numbers

• First, consider two denary numbers with different
mantissas, 1534 x 103 and 1025 x 102
• Clearly we cannot simply add 1534 and 1025, since
these numbers represent 1,534,000 and 102,500
• Similarly, we cannot add the mantissas of two
floating point binary numbers until we have
equalised the mantissas
• To equalise the mantissas, convert the numbers to
fixed point binary
Floating point arithmetic
Unit 6 Data Types

Example
• Add the two floating point numbers
A = 0.1100000 0001 B = 0.1111100 0011
• Convert to fixed point and add:
A = 1.1000
B = 111.1100
Sum = 1001.0100
• The number is positive, so sign bit will be 0
• Convert to normalised floating point:
Result = 0.1001010 0100
Floating point arithmetic
Unit 6 Data Types

Subtracting floating point

numbers
• To subtract a floating point number from another, first
convert them both to fixed point
• Find the two’s complement of the number to be
subtracted
• Add the two numbers
• Convert result to normalised floating point
Floating point arithmetic
Unit 6 Data Types

Example
• Subtract the two floating point numbers, A - B
A = 0.1101000 0100 B = 0.1110000 0011
• Convert to fixed point:
A = 1101.0000
B = - 0111.0000 (one’s complement =
1000.1111)

• Find two’s complement of B: 1001.0000

• Add A + (-B): 0110.0000 (ignore overflow bit)
• Convert back to normalised floating point:
Result = 0.1100000 0011
Floating point arithmetic
Unit 6 Data Types

Worksheet 4
• Now try Task 3 on the worksheet
Floating point arithmetic
Unit 6 Data Types

Plenary
• Be sure you can:
• Convert floating binary numbers to decimal and vice versa
• Normalise floating-point numbers with positive and negative
mantissas
• Explain the effect of increasing the size of the mantissa and
exponent
• Add and subtract floating point numbers
Floating point arithmetic
Unit 6 Data Types

Copyright

© 2016 PG Online Limited

The contents of this unit are protected by copyright.

This unit and all the worksheets, PowerPoint presentations, teaching guides and other associated files
distributed with it are supplied to you by PG Online Limited under licence and may be used and copied by you
only in accordance with the terms of the licence. Except as expressly permitted by the licence, no part of the
materials distributed with this unit may be used, reproduced, stored in a retrieval system, or transmitted, in any
form or by any means, electronic or otherwise, without the prior written permission of PG Online Limited.

Licence agreement

This is a legal agreement between you, the end user, and PG Online Limited. This unit and all the worksheets,
PowerPoint presentations, teaching guides and other associated files distributed with it is licensed, not sold, to
you by PG Online Limited for use under the terms of the licence.

The materials distributed with this unit may be freely copied and used by members of a single institution on a
single site only. You are not permitted to share in any way any of the materials or part of the materials with any
third party, including users on another site or individuals who are members of a separate institution. You
acknowledge that the materials must remain with you, the licencing institution, and no part of the materials may
be transferred to another institution. You also agree not to procure, authorise, encourage, facilitate or enable any
third party to reproduce these materials in whole or in part without the prior permission of PG Online Limited.