COMPUTER

ARITHMETIC
Electronics Engineering Department
Faculty of Engineering
University of Santo Tomas

COMP421 UST-ECE
©engradc 2010
©engrkad 2013
Different Data Types
 Unsigned Notations
 Unsigned Non-negative
 Unsigned 2’s complement

 Unpacked and Packed Binary Coded Decimal

(BCD)
 Single and Double Precision Floating Point
Format
 Alphanumeric Codes
 Unsigned Non-Negative
 Binary numbers are always treated as positive
numbers
 For 8-bit binary number

27 26 25 24 23 22 21 20

 For 16-bit binary number

215 214 213 212 211 210 29 28 27 26 25 24 23 22 21 20

 Unsigned Two’s Complement
 MSB of the signed binary number is the sign bit
 indicates whether the number is positive or negative
 For 8-bit binary number

-27 26 25 24 23 22 21 20

 For 16-bit binary number

-215 214 213 212 211 210 29 28 27 26 25 24 23 22 21 20

Unsigned Notation

Two’s
BINARY Non-Negative
Complement

0000 0000 0 0
0111 1111 127 127
1000 0000 128 -128
1000 0001 129 -127
1111 1111 255 -1
Arithmetic Operation

 Examples:
a. (+7) + (+4)
b. (+15) + (-6)
c. (+16) + (-24)
d. (-5) + (-9)
e. (+9) – (+3)

 Unsigned integers cannot represent negative numbers. A negative sign on the

number implicitly defines a subtraction operation.
 To perform subtraction, perform 2’s complement addition on the
subtrahend.
 Non-Negative Addition

1 1 1111 111
7 d 0000 0111 b 255 d 1111 1111b
+ 4 d 0000 0100 b + 1d 0000 0001 b
11 d 0000 1011 b 256 d 0000 0000 b
=0
out of range
1 1 111 11
127 d 0111 1111 b 127 d 0111 1111 b
+ 128 d 1000 0000 b + 150 d 1001 0110 b
255 d 1111 1111 b 277 d 0001 0101 b
out of range = 21

OVERFLOW: Carry Out = 1

Non-Negative Subtraction

XX–Y
X  X + ( –Y)
1 1111 11
2d 0000 0010 b 0000 0010 b
– 1d – 0000 0001 b + 1111 1111 b
1d 0000 0001 b
1 1111 11
15 d 0000 1111 b 0000 1111 b
– 6d – 0000 0110 b + 1111 1010 b
9d 0000 1001 b
Non-Negative Subtraction
XX–Y
X  X + ( –Y)
1d 0000 0001 b 0000 0001 b
– 2d – 0000 0010 b + 1111 1110 b
-1 d out of range
1111 1111 b
= 255
wrong answer

6d 0000 0110 b 0000 0110 b

– 15 d – 0000 1111 b + 1111 0001 b
-9d 1111 0111 b
out of range
= 247
wrong answer
OVERFLOW: Carry Out = 0
 Two’s Complement Addition

1 0 1111 111
7d 0000 0111 b 127 d 0111 1111b
+ 4d 0000 0100 b + 1d 0000 0001 b
11 d 0000 1011 b 128 d 1000 0000 b
out of range = -128
wrong answer
1 1111 111 1
-127 d 1000 0001 b -128 d 1000 0000 b
+ -1 d 1111 1111 b + -1 d 1111 1111 b
-128 d 1000 0000 b -129 d 0111 1111 b
out of range = 127
wrong answer
OVERFLOW: Carry Out ≠ Carry In (to MSB)
Two’s Complement Subtraction
XX–Y
X  X + ( –Y)
1 1111 11
2d 0000 0010 b
– 1d + 1111 1111 b
1d 0000 0001 b
1 1111 11
15 d 0000 1111 b
– 6d + 1111 1010 b
9d 0000 1001 b
Two’s Complement Subtraction
XX–Y
X  X + ( –Y) OVERFLOW:
0 0000 000 Carry Out ≠
1d 0000 0001 b Carry In (to MSB)
– 2d + 1111 1110 b
-1 d 1111 1111 b
Valid answer = -1
1 0000 000
-128 d 1000 0000 b
– 1d + 1111 1111 b
-129 d 0111 1111 b
= 127 Wrong answer
out of range
Binary Adder/Subtractor Implementation

X Y (1’s comp Y)

0 1
MUX Select ?

Cout Parallel Adder Cin

Overflow in Non-Negative Integer

 Addition Process
 ADD: XX+Y
 Arithmetic overflow
 The result of addition process is out of range
 Setting an overflow flag

 Theresult must be in the range: 0 to 2n – 1

Example:
1111 1111b + 0000 0001b = 1 0000 0000b

 Therefore: Overflow occurs when the carry out is 1

Overflow in Non-Negative Integer
 Subtraction Process
 SUB: X  X – Y or X  X + Y’ + 1
 Overflow occurs when the result is less than zero
 Examples:
2–1=1
255 – 254 = 1
1 – 2 = –1
254 – 255 = –1

 Therefore:
Overflow occurs when the carry out is 0
Overflow in 2’s Complement Integer

 Addition / Subtraction Process

 ADD: XX+Y SUB: X  X + Y’ + 1
 The result must be in the range: –2n –1 to 2n –1 – 1
 Overflow only occurs when two numbers with the same
sign are added.
 Examples:
126 + 1 = 127
–127 + (–1) = –128
127 + 1 = 128
–128 + (–1) = –129
 Therefore: Overflow occurs when the carry out
and the carry in to MSBs are not equal
Unpacked and Packed BCD Format
 BCD is used in computers to store decimal data

Decimal
Unpacked BCD Packed BCD
Number

18 00000001 00001000 0001 1000

905 00001001 00000000 00000101 1001 0000 0101
BCD Addition
 BCD sum cannot have digit with a value greater
than 9.
 Therefore, a carry must be generated if the
result of addition is greater than 9 (or 1001).

 Examples:
a) 35 + 23 c) 35 + 66
b) 7 + 5 d) 8 + 9
BCD Addition
01 111
35 d 0011 b 0101 b
+ 23 d 0010 b 0011 b
58 d 0101 b 1000 b
5 8
BCD Addition
1 111
7d 0000 b 0111 b
+ 5d + 0000 b + 0101 b
11
12 d 0001 b 1100 b INVALID BCD
+ 0110 b Correction Factor

VALID BCD 0010 b

BCD Addition

1 1 11 1
35 d 0011 b 0101 b
+ 66 d + 0110 b + 0110 b
111 111
101 d 1010 b 1011 b
+ 0110 b + 0110 b
0000 b 0001 b
BCD Addition
1
8d 1000 b
+ 9d + 1001 b WRONG RESULT!
17 d 0001 b VALID BCD
+ 0110 b Correction Factor

0111 b
BCD Addition
 2 possible errors:
 The sum is an invalid BCD code (greater than 9)
 The sum is a valid BCD code (0-9) but not the correct
result
 Example:

7 +5
 35 + 66
8 + 9
 Solution:
 Add a correction factor (0110)
 BCD Adder Implementation
X Y
4 4

Cout1 Cin
Parallel Adder 0110 0000
4 4

4 1 0
MUX Select ?
Cout
4

Cout2
Parallel Adder Cin = 0

4
X+Y
BCD Subtraction

 Examples:
a) 9 – 6 c) 17 – 9
b) 25 – 11 d) 41 – 18
BCD Subtraction
0 1
9d 1001 b
– 6d – 0110 b
3d 0 011 b
3

0
25 d 0010 b 0101 b
– 11 d – 0001 b – 0001 b
14 d 0 0 01 b 010 0 b
1 4
BCD Subtraction
0
17 d 0001 0111 b
– 9d – 0000 1001 b
8d 0000 111 0 b INVALID BCD
0 – 0110 b Correction Factor
1 0 00 b
8
BCD Subtraction
01 1
41 d 0100 0001 b
– 18 d – 0001 1000 b
01
23 d
0010 1 0 01 b VALID but WRONG
2 – 0110 b Correction Factor

0 011 b
3
BCD Subtraction
 If the initial difference contains BCD digit greater
than 9 (or 1001) or if a borrow from the next
higher digit was required to do subtraction then,
a correction factor of 6 (or 0110) must be
subtracted from the initial difference.

 Now, we already know how to perform BCD

subtraction. But, how do we implement it in our
design?
BCD Subtractor Implementation
X 9’s comp Y

Parallel Adder Cin = 1

Cout1

0110

Cout2 Parallel Adder Cin = 0

(discarded)

X–Y
BCD Subtractor Implementation
X 9’s comp Y
Cin = 1

0110 0000
Parallel Adder
Cout1
1 0
MUX Select ?

Cout2 Parallel Adder Cin = 0

X–Y
Floating Point Binary Number
 Represents numbers in scientific notation and is capable of
representing large and small numbers without an increase
in the number of bits.
 Composed of sign, significand (or mantissa) and exponent
[sign]1.[significand]2 x 2 [exponent]

 Types
 Single Precision Floating Point
 Double Precision Floating Point
Single Precision Floating Point
Binary Number
 Uses 32 bits
 Has a sign bit, 8-bit biased exponent and a 24-bit
mantissa (but only the 23-bit fractional part of it is
presented, it contains 1 implied hidden bit)
 Sign bit = 1 if negative, 0 if positive
 8-bit biased exponent = 127 + exponent of the
normalized real number
 Single Precision Floating Point
Binary Number
 Single Precision Floating Point Format
B B B B B
31 30 23 22 0

Sign
Bit Biased Exponent Significand or Mantissa
Double Precision Floating Point Binary
Number
 Uses 64 bits
 Has a sign bit, 11-bit biased exponent and a 53-bit
mantissa (but only the 52-bit fractional part of it is
presented, it contains 1 implied hidden bit)
 Sign bit = 1 if negative, 0 if positive
 11-bit biased exponent = 1023 + exponent of the
normalized real number
 Double Precision Floating Point
Binary Number
 Double Precision Floating Point Format
B B B B B
63 62 52 51 0

Sign
Bit Biased Exponent Significand or Mantissa
Single and Double Precision:
COMPARISON
 Single precision
SIGN EXPONENT (8 bits) MANTISSA (23 bits)

31 30 23 22 0

 Double precision
EXPONENT (11 bits)

SIGN MANTISSA (52 bits)

63 62 52 51 0
Floating Point Binary Number
 Steps in converting decimal number to its floating
point equivalent
 Convert the decimal number to binary
 Express the binary number into its normalized form
(in scientific notation)
 Determine the binary value of the sign bit and biased
exponent
 Arrange the sign bit, biased exponent and fractional
part of the mantissa.
Example
1. Represent the following decimal numbers in single
precision floating point format
a. +12.25
b. – 0.75
c. –45.5625
2. Represent the following decimal number in double
precision floating point format
a. +519
b. –314.1592
American Standard Codes for Information
Interchange (ASCII)
ASCII
 0 – 9  011 XXXX
 A – Z  100 XXXX to 101 XXXX
 a – z  110 XXXX to 111 XXXX