Data Manipulation

1
Data Manipulation

• In a major matter, no details are small.
• - French Proverb

2
Computer Architecture

• Typical components of a computer system
• Central Processing Unit (CPU)
• Primary Storage (Main Memory)
• Secondary Storage
• I/O Devices
• System Bus

3
Components of CPU
4
CPU Components

• ALU Arithmetic Logic Unit
• Circuits for arithmetic and logic data
  manipulation
• CU Control Unit
• Circuits for coordinating and controlling
  activities
• Registers
• General-purpose Registers storing intermediate
  results temporarily for CPUs use
• Special-purpose registers storing instructions
  and program counter

5
Primary Storage

• Known as Main Memory
• Sequence of contiguous, adjacent cells
• with unique address corresponding to its physical
  location in the sequence
• Typically random-access (RAM)
• Random-access or direct-access allows CPU to
  access a specific address location in memory for
  reading or writing

6
Main Memory

• Storage area for bits
• Bits are grouped into (typically 8-bit) chunks
  called bytes
• Bytes are grouped into (typically 4-byte) chunks
  called words
• Each word has a memory address
• Memory capacity is measured in bytes example,
  256 MegaBytes of RAM

7
Main Memory Word Size

• Word is loosely defined as the amount of data
  that CPU processes at one time
• Word Size usually matches the size of CPU
  registers
• Typically, Word Size is 32-bits in modern
  computers
• Word Size is a fundamental CPU design decision
  that has implications for the design of other
  components like the Bus width

8
Processor Registers

• Temporary storage areas in CPU
• Two categories
• General purpose
• Special purpose
• Total number of registers in a CPU is variable by
  design
• Each register is referenced by a unique register
  number
• For example, in 4-bit notation, register numbers
  would range from 00002 to 11112, a total of 16
  registers
• For purposes of notation, register numbers and
  memory addresses usually are expressed in Hex
• So, 16 registers in Hex would be numbered 016
  through F16

9
General purpose registers

• Used in a variety of ways by programs
• Hold data, intermediate results and such that are
  frequently accessed during execution, rather than
  fetch from main memory every time
• Increase in number of general-purpose registers
  would increase execution speed of a program
• Unfortunately, registers are very expensive, so
  we need to establish a trade-off

10
Special purpose registers

• Content and use are specified as part of CPU
  design
• Two of the typical special-purpose registers are
• The Instruction Register
• Used by Control Unit to hold instruction just
  loaded from memory, before decode and execute
• The Program Counter (or, instruction pointer)
• Pointer to the next instruction to be executed by
  CPU

11
System Bus

• A set of parallel communication lines connecting
  CPU with main memory and other system components
• Dedicated buses control bus, address bus, data
  bus

12
Summary CPU Architecture
13
Secondary Storage

• Before we talk about Machine Language and Machine
  Instructions, let us take a brief look at
  Secondary or Mass Storage devices.
• (Refer Chapter 1, Section 1.3)

14
Memory Organization

• Memory Hierarchy
• in
• modern computers

15
Library Example

• Common known case book_1 and book_2 are most
  commonly used references in this library. You can
  borrow only one of them at a time.
• You go to the library to borrow book_1. It is
  stored way back in some dark, hard-to-access
  corner
• You manage to find and get book_1 using
  card-catalog and are now done with it.
• Now you want to get book_2
• Wouldnt it be easier if book_2 was located right
  next to book_1 rather than having to search all
  over the library?

16
Memory and Storage

• Basic Principles
• Make the common case fast
• Temporal and spatial locality
• Secondary storage contains all the data and
  instructions not currently needed by CPU
• It should be easy to access, should hold data
  indefinitely, and should be both read-able and
  write-able

17
Basic Memory Organization

• Registers (CPU)
• General-purpose, Special-purpose
• High-speed, quick response times
• Cache
• Often located within CPU
• High-speed response time similar to CPU Registers
• Holds a copy of that potion of Main Memory in
  current use
• Main Memory
• Collection of adjacent cells
• Accessed via system bus by CPU
• Secondary Memory, I/O devices

18
Main Memory

• Memory cells are usually byte-sized and arranged
  by their addresses, so, easy for random access
  (RAM)
• If you know the cell address, you can get the
  contents of that cell

19
Byte-size memory cell
There is no physics demarcation for cell
boundaries. Chunks of 8 bits forms an imaginary
grouping for our convenience. Data stored in a
memory cell is in the form of 0s and 1s, which
we envision as arranged in a row, left-to-right.
Figure 1.7, Page 27The organization of a
byte-size memory cell
20
Memory Capacity Units

• One byte 8 bits
• 210 bytes 1024 bytes 1 kilo-byte
• 220 bytes 1,048,576 bytes 1mega-byte
• 230 1 giga-byte

21
Secondary Storage Devices

• On-line storage
• Devices that are connected and readily available
  to the machine without human intervention
• Off-line storage
• Devices that need human intervention to make them
  available to the machine
• Mass Storage Devices
• can be on-line, or, off-line devices
• with direct access or sequential access
• Examples
• Magnetic Disks (direct access)
• Compact Disks (direct access)
• Magnetic Tapes (sequential access)

22
Magnetic Disks Hard Drive
23
Hard Disk Drive

• Refers to magnetic disks with hard metal
  substrates (bases)
• Disk or Platter size is about 14 in diameter
• Several platters mounted on disk drive devices
• Data is recorded in concentric tracks
• Read/Write heads do the actual data access
• Capacity ranges from 50 MB to 1.5 GB
• Direct access device
• Performance evaluated in terms of access time and
  transfer rate

24
Disk Storage System
Figure 1.9,Page 31 A disk storage system

• Access Time
• Seek time Rotational delay
• Seek Time time required to position the
  read/write head along the correct track
• Rotational Delay time required to rotate the
  platter to position the correct sector underneath
  the read/write head

25
Magnetic Tape Storage
Figure 1.11 Page 33A magnetic tape storage
mechanism

• Like an audio tape, this is a sequential access
  device
• Data must be accessed in the physical order in
  which they were stored in the tape
• Very slow, but appropriate and cost-effective for
  archival copies for low-demand applications
• Capacity measured in terms of recording density
  Bytes per inch , Bpi
• Max recording density can range as high as 36,000
  Bpi

26
Compact Disks

• Like LPs, but spiraling inside to outside
• Has tracks and sectors just like magnetic disks
• Laser beam reads the data, just like read/write
  head does in magnetic disks
• Typical capacity 600MB to 700 MB

27
Logical vs. Physical Records
Figure 1.12Logical records versus physical
records on a disk

• File Storage/Retrieval Exercise remember writing
  your first name, last name, favorite sports/games
  on separate sheets of paper a while ago?
• Physical Records are on Paper1, Paper2 and Paper3
• How can I get the Logical Records
• First name, Last name, favorite sports/games per
  student
• Logical record size need not match the physical
  record size of the disk

28
Summary Memory Hierarchy
29
Storage Options at a Glance
30
Summary Memory Access

• Memory of a computer stores data and program for
  processing
• Memory Addressing
• Organizing as a sequence of addressable cells,
  each with a unique address
• Memory Allocation
• Assignment of specific memory address to data and
  other elements of system software

31
Back to Data Manipulation

• Well try to understand certain terminology we
  will be using from here on

32
Clock rate

• If CPU is the heart of a computer system, then
  clock-rate is like its heartbeat.
• System clock generates timing pulses, clock
  ticks, and sends it over separate line on control
  bus to all other components of the computer
• CPU actions are timed according to this system
  clock
• CPU is designed to do several actions in one tick
  of this system clock
• Typical CPU actions are fetch instruction or
  data and execute instruction

33
Clock Rate Megahertz (MHz)

• Definition number of cycles per second is known
  as the frequency, expressed in units of Hertz
  after the physicist Hertz
• Frequency of the system clock is measured in
  millions of cycles per second, or, Megahertz, MHz
  for short.
• Typical desktops have a few hundred MHz clock
  speed
• Typically, CPUs performance is measured in terms
  of Millions of Instructions per Second (MIPS)

34
Measuring CPU Performance

• So, is a 500 MHz machine better than a 200 MHz
  machine?
• Cannot compare unless we specify they are the
  same CPU architecture design
• Total number of clock cycles needed to complete
  an instruction differs between CPU designs, so
  clock speed alone does not help in comparing
  performance
• A meaningful comparison is via Benchmarking
  like, comparing how different machines execute
  the same program

35
Stored Program Concept

• Early computers were designed to perform a single
  task at a time and needed to be re-wired for
  different tasks
• Code breaking
• Artillery ballistics
• The instructions were hard-wired
• More like a music box that plays the same music
  every time
• To be more useful, needed the flexibility of a
  CD-changer

36
Program vs. Data

• Flexibility in early computers was achieved by
  realizing that programs (i.e., instructions to
  manipulate data) can be treated the same way as
  data can be treated
• Programs can be encoded and stored in memory,
  just like data
• Rather than fetching just data and executing the
  hard-wired instructions, store both instructions
  and data and fetch them both as needed to execute
  the task

37
Von Neumann Machines

• Though credited to von Neumann, stored-program
  concept was apparently developed by researchers
  at J.P Eckert and Moore School of Electrical
  Engineering, UPenn.
• Essential features
• Single general-purpose processor
• Stored programs
• Sequential processing of instructions
• Alternating instruction and execution cycle

38
Back to Data Manipulation

• A closer look at what happens inside a machine

39
Machine Language Objectives

• Machine Language
• Instruction Set architecture
• RISC, CISC
• Op-code, Operands
• Machine Cycle
• Von Neumann Machine Primitives
• Data-Copying (Transfer) operations
• Data Transformation operations
• Sequence Control

40
Machine Language

• Instruction Set is the basic set of operations
  that a computer can perform
• Processor hardware implements a pre-defined
  instruction set
• Any operation not defined in the processor
  instruction set must be implemented in software
  as a sequence of these pre-defined processor
  instruction set
• Fundamental design trade-off size and
  complexity of the instruction set

41
Basic Instructions

• Data Transfer
• copy data from one location in memory to another
• Store data in a specified location in memory
• Examples Load, Store, Move
• Involves registers, main memory, system bus
• Data Transformation
• Perform arithmetic/logic operations on data
• Examples AND, OR, XOR, ADD, SHIFT, ROTATE
• Involves ALU, registers
• Sequence Control
• Change the order in which instructions are
  executed
• Dont really deal with data manipulation
  directly, but rather with instruction execution
• Examples Branch (JUMP), Halt
• Involves special-purpose registers, control unit

42
Machine Instructions CISC

• CISC Complex Instruction Set Computing
• Early machines had limited memory as memory was
  very expensive, so had to keep programs short
• One response to memory issues is Design complex
  set of instructions that combine several simpler
  instructions
• Example, Floating-point arithmetic rather than
  handling whole and fraction parts separately and
  combining the result using a sequence of simple
  operations, implement a complex operation
• Such implementations might tend to be slow,
  depending on their design
• Intels Pentium series of processors use CISC

43
Machine Instructions RISC

• RISC Reduced Instruction Set Computing
• No complex instructions
• Fixed-length instructions that are shorter than
  CISC
• Reduced circuitry, compared to CISC, means lower
  cycle times, therefore higher speeds
• PowerPC series of processors use RISC
• Uses Registers for data transformations rather
  than rely on main memory

44
Trade-offs RISC vs. CISC

• CISC
• Typically, combines data copying and data
  transformation operations in the same instruction
• Complex instructions implemented as micro program
• Example Add(addr1, addr2, addr3)
• Simply means
• Get contents from memory addr1
• Get contents from memory addr2
• Add the two contents
• Store the result in memory addr3
• One instruction does it all!
• Advantage
• Might use only one processor cycle to complete
  executing the above instruction
• Other similar processor cycle savings possible in
  other instructions
• Saves memory space

• RISC
• Separate data copying and data transformation
  instructions
• Simple instructions implemented in hardware
• Same CISC example
• LOAD addr1
• LOAD addr2
• ADD contents of addr1 and addr2
• STORE result in addr3
• FOUR simple instructions to get the same job done
• Advantage
• Simple circuitry, means smaller physical
  machine-size
• Lower cycle times, means higher speeds
• Uses registers which have faster response time
  than accessing main memory via bus

45
Hypothetical Question

• Question Which of the two CPU Instruction set
  designs (RISC or CISC) would take more time to
  finish the same task that of adding two numbers
  and storing the result?
• Assume
• a RISC processor cycle takes 1 second to complete
  each one of its instructions
• a CISC processor also takes 1 second to complete
  each one of its instructions
• one CISC instruction as in the example we just
  saw takes one processor cycle to complete
• viz., Add(addr1, addr2, addr3)
• each one of the RISC instructions as in the
  example we just saw in previous slide takes the
  one processor cycle to complete
• viz., LOAD addr1, LOAD addr2, ADD (addr1,
  addr2), STORE addr3

46
Summary Instruction Set Architecture

• Cannot compare performance without knowing the
  specific CPU Instruction Set
• CISC uses less memory than RISC but involves
  complex circuitry
• RISC uses simpler circuitry than CISC but more
  memory
• Is program execution faster with complex
  instructions but relatively slow processor ?
• Or, is program execution faster with simpler
  instructions with a faster processor?

47
Machine Instruction Format

• Just like data and programs, an instruction is
  represented as a string of bits (0s and 1s)
• Different computers use different format to
  specify machine instructions
• Each instruction consists of two parts
• Op-code, short for operation-code
• Operands
• Op-code indicates which operation to perform
  like AND, XOR, STORE, LOAD, SHIFT
• Operand indicates all supplementary details
  needed for executing the op-code like the data
  location, or data value

48
Machine Cycle Fetch-Decode-Execute

• Machine Cycle is a three-step process that goes
  on and on until instructed to Stop
• Fetch
• Decode
• Execute

49
Machine Cycle Fetch-Decode

• Fetch step
• CU asks Main memory to give the instruction
  indicated in Program Counter, and increments the
  program counter
• CU puts the current instruction in Instruction
  Register
• Decode step
• CU separates the instruction in the Instruction
  Register to op-code and operands
• CU internally signals ALU to start executing

50
Machine Cycle Execute

• Execute step
• CU activates appropriate logic circuitry of ALU
  to execute the instruction
• Data passes through the circuitry and produces an
  output (result) for that instruction
• The result is placed in a register or CU writes
  it to main memory
• CU again starts the Fetch step!
• Remember that CU already incremented the Program
  Counter

51
Machine cycle Flow of control
52
Brookshear Machine

• 16 general-purpose registers
• numbered 0 through 15
• Hex notation numbered 0 through F
• 256 byte-size main memory cells (i.e., 8 bits
  each)
• numbered 0 through 255
• Hex notation numbered 00 through FF
• 12 simple instructions
• encoded using 16 bits (2 bytes) per instruction
• Hex notation 4 hex digits per instruction
• One hex digit for op-code
• Other three hex digits for Operands

53
Test Your Knowledge

• The Brookshear machine instruction provides 4
  bits to identify a general purpose register
• How many registers can the machine have?
• How many bits in a Brookshear byte?
• The Brookshear machine instruction provides for a
  byte-size memory address that addresses
• How many memory locations are addressable
  directly?

54
Test Your Knowledge Answers

• 4 bit register addresses
• Maximum combination is
• 2 x 2 x 2 x 2 24 16 possible registers
• 8 bits in a Brookshear byte
• 1 byte memory address
• Brookshear byte is 8 bits
• Maximum combination is
• 28 256 possible memory locations

55
Instruction format for Brookshear machine
56
Brookshear Instruction Example
Actual 16-bit patterns per instruction
0011
0101
1010
0111
Hex form (4 bits)
3
5
A
7
Operand
Op-code
57
Brookshear Machine Architecture
Figure 2.4, Page 79 The architecture of the
machine described in Appendix C
58
Brookshear machine Appendix C

• Study Appendix C and do the following exercise
  (5 minutes)
• How many Data Copying/Transfer instructions are
  encoded for this machine?
• How many Data Transformation (i.e.,
  Arithmetic/Logic) instructions are encoded for
  this machine?
• How many Control instructions are encoded for
  this machine?
• Complete the table by listing the Brookshear
  machine instructions in the appropriate columns

59
Brookshear machine Appendix C

• Answers
• FOUR Data Copying/Transfer instructions are
  encoded for this machine
• SIX Data Transformation (i.e., Arithmetic/Logic)
  instructions are encoded for this machine
• TWO Control instructions are encoded for this
  machine
• Complete the table by listing the Brookshear
  machine instructions in the appropriate columns

Study the difference between LOAD1 and
LOAD2 ADD1 and ADD2
60
Exercise trace by hand

• Using Appendix C, trace the following machine
  instruction by hand and write the results, given
  the contents of main memory as in the table
• 156C
• 166D
• 5056
• 306E
• C000

Main memory
61
Program Execution Examples

• Finally you get to use a computer!!
• Boot up your machine and open a browser to the
  following address
• http//cs01.pcc.edu/eodekirk/cs160/bsmachine/mach
  inegui.html

62
Brookshear Machine Simulator
63
Sample Program
64
Program 1
65
Communicating with other devices
Figure 2.13 Controllers attached to a machines
bus