3' Basic Architecture

1
3. Basic Architecture Information Representation

• Basic computer architecture.
• Information representation.
• The Arabic number system.
• Binary, Octal and Hexadecimal.
• ASCII, EBCDIC and Unicode.
• Burrell 2.1, 2.2, 2.3, 2.7.
• Clements 1, 4 (intro) 4.1, 4.2, 4.3 (intro)
  4.3.1.

2
Basic Computer Architecture

• CPU (Central Processing Unit) Does the work.
• Memory Stores data.
• I/O Unit Communicates with the world.
• Bus Carries data around the computer.
• Lots of buses inside the functional units as
  well. Later.

3
Information Representation

• Computers process information represented using
  patterns of 0s and 1s
• Binary digits (bits).
• Stored in memory by 2-state electronic devices
  (bistables) and on I/O devices by 2-state
  magnetic devices (usually).
• On modern Complementary Metal Oxide Semiconductor
  (CMOS) chips 0 is represented by -15 Volts and 1
  by 15 Volts (although -12 and 12 are sometimes
  used).
• On Transistor Transistor Logic (TTL) chips 0 is
  represented by 0 Volts and 1 by 5 Volts.
• How can patterns of bits represent information?
• Numbers, text, pictures, sound etc.

4
The Arabic System

• Humans have used lots of different ways of
  representing numbers
• Caveman
• I II III IV V Roman
• 0 1 2 3 4 5 Arabic
• Finally seem to have settled on the Arabic
  system.
• In this system the relative position of the
  symbols is important
• 123 (1 102) (2 101) (3 100)
• 321 (3 102) (2 101) (1 100)
• 10 is the base of the Arabic system that we use.
  Called Base 10 or Decimal.

5
Binary

• Can use an Arabic system with any base.
• Computers use Base 2 (Binary) because its easy
  to build electronic bistables.
• 010101012 (0 27) (1 26) (0 25) (1
  24) (0 23) (1 22) (0 21) (1
  20)
• 85
• Convention numbers without suffixes are base 10.

6
Binary II
7
Octal Hexadecimal

• Computers usually group patterns in sequences of
  8, 16, 32 or 64 bits (bytes and words).
• Takes too long to write out all the bits. Too
  easy to make mistakes.
• Instead collect the bits in patterns of 3
  (Octal) or 4 (Hexadecimal), starting at the right
  hand end and write down the decimal digit which
  represents the same number as the pattern
  represents in binary.
• Binary 0101 0101
• Hexadecimal 5 5
• Binary 001 010 101
• Octal 1 2 5

0s added at front to make up the numbers.
8
Octits Hexits

• In Octal we must represent patterns of 3 bits.
• Need symbols for 0 (0002) to 7 (1112).
• Use corresponding decimal digits.
• In Hexadecimal we must represent patterns of 4
  bits.
• Need symbols for 0 (00002) to 15 (11112).
• Use corresponding decimal digits.
• No single decimal digits to represent numbers
  from 10 to 15 so we use letters of the alphabet
• Decimal 10 11 12 13 14 15
• Hexadecimal A16 B16 C16 D16 E16 F16

9
Octal Û BinaryÛ Hexadecimal

• Octal º Base 8. Hexadecimal º Base 16.
• Easy way to convert between them
• 1778 001 111 1112
• 0111 11112
• 7F16
• 5516 0101 01012
• 001 010 1012
• 1258
• Works because 8 23 and 16 24 so 3 bits can
  be swapped for 1 octit and 4 bits can be swapped
  for 1 hexit.

10
Decimal Û Binary Conversion

• Octal Û binary Û hexadecimal is easy
• Conversion from binary to decimal by repeated
  multiplication.
• Or via method on slide 3.6.
• Conversion from decimal to binary by repeated
  division
• 2 13
• ---
• 2 6 1 1 20
• ---
• 2 3 0 0 21
• ---
• 2 1 1 1 22
• ---
• 0 1 1 23

Read off remainders in reverse order to obtain
result 11012
11
Bits, Characters Text

• People dont converse in bit patterns. They
  converse in sequences of characters
• A truly wise man does not play leapfrog with a
  Unicorn.
• A computer must be able to represent sequences of
  characters.
• A computer must be able to represent characters.
• A pattern of 8 bits is called a byte.
• Each byte can be used to represent a character.

12
ASCII EBCDIC

• ASCII American Standard Code for Information
  Interchange
• Character Decimal Binary
• A 65 1000001
• 32 0100000
• t 116 1110100
• r 114 1110010
• Some old IBM ICL computers use EBCDIC
  Extended Binary Coded Decimal Interchange Code.
• Note A byte is 8 bits but ASCII only uses 7
  bits for each character. What is the 8th bit for?
• Parity checking (RTFM).
• ISO 8859-1 Latin code (extended ASCII).

13
Unicode

• ASCII is an American invention.
• Only includes characters for American English.
• No accents, only Latin alphabet.
• No sign!
• Unicode is an international code that has
  recently been standardised.
• 16 bits Þ 65536 character codes available.
• Has accents, Greek, Chinese, Japanese, Cyrillic
  alphabets.
• Has sign!
• First 128 characters of Unicode are the same as
  ASCII.
• First 256 are the same as extended ASCII.
• Most modern machines support Unicode in hardware.
  
• Some modern languages (e.g. Haskell, Java)
  support it in software.

14
Summary

• Essentially, a computer consists of CPU, Memory
  and I/O functional unit(s) connected together via
  buses.
• Computers process information represented as
  patterns of 0s and 1s (patterns of bits).
• Humans represent numbers via the Arabic system
  using base 10 (decimal or denary).
• Computers represent numbers via the Arabic system
  using base 2 (binary).
• Bit patterns are interpreted as if they were
  binary numbers.

15
Summary II

• Humans often use Arabic base 8 (Octal) or Arabic
  base 16 (Hexadecimal) to represent bit patterns
  as this reduces mistakes.
• Conversion between Octal or Hexadecimal and
  binary is by bit grouping.
• Conversion between Octal and Hex is via binary.
• Conversion between decimal and binary is by
  repeated division or multiplication (nb
  shortcut on slide 3.6).

16
Summary III

• A pattern of 8 bits is called a byte.
• A byte can be used to represent a character.
• Usual encoding is the 7 bit ASCII (American
  Standard Code for Information Interchange).
• Some machines use 8-bit extended ASCII.
• Much less common is the 8 bit EBCDIC (Extended
  Binary Coded Decimal Interchange Code).
• Catching on is the 16 bit Unicode.
• A two byte code.
• A sequence of bytes can represent text.