2. General System Architecture

Embed Size (px)

Citation preview

  • 8/7/2019 2. General System Architecture

    1/28

    ECE V Semester

    COMPUTER ARCHITECTURE&

    ORGANIZATIONUnit 2:

    General System Architecture

    MANAV

    RACHNA

    COLL

    EGEOFENGINEERI

    NG

  • 8/7/2019 2. General System Architecture

    2/28

    y

    Contents

    Instruction Codes

    Stored Program Concept

    Flynns ClassificationSISDSIMDMISD

    MIMDMulti level Viewpoint of a machine

    Structured Organization

    MANAV

    RACH

    NA

    COLL

    EGEOFENGINE

    ERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    3/28

    Instruction Codes

    Internal Organization of a digital system is defined by thesequence of micro operations it performs on data stored in itsregisters.

    The general purpose digital computer is capable of:Executing various micro operations, andCan be instructed as to what specific sequence of operations it

    must performThe user controls the process by means of a program.

    A Program is a set of instructions that specify the operations,operands, and the sequence by which processing has to occur.

    A computer instruction is a binary code that specifies a sequenceof micro operations for the computer. Instruction codes together with data are stored in memory.

    MANAV

    RACH

    NA

    COLL

    EGEOFENGINE

    ERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    4/28

    Instruction Codes (Cont)

    Instruction Codes : A group of bits that instructthe computer to perform a specific operation.

    It is usually divided into parts, each having its ownparticular interpretation.

    Most basic part of an instruction code is itsOPERATION part.

    Opcode : A group of bits that define operations

    like ADD, SUBTRACT etc.The number of bits required for the operation codeof an instruction depends on the total number ofoperations available in the computer.

    The opcode must consist of atleast n bits for a

    given 2n (or less) distinct operations.

    MANAV

    RACH

    NA

    COLL

    EGEOFENGINE

    ERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    5/28

    Stored Program Concept

    The simplest way to organize a computer is to haveone processor register & an instruction code with 2parts:Opcode (What operation is to be completed)Address (Address of the operands on which the

    operation is to be computed)

    A computer that by design includes an instructionset architecture and can store in memory a set ofinstructions (a program) that details thecomputation and the data on which computation isto be done.

    MANAV

    RACHNA

    COLL

    EGEOFENGINE

    ERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    6/28

    Stored Program Concept (Cont)

    MANAV

    RACHNACOLL

    EGEOFENGINE

    ERING

    Pre ared B : Nidhi Gar

    Opcode

    Processor Register(Accumulator or AC)

    Address

    Binary OperandOperands

    (Data)

    Instructions

    (Program)

    15 12 11 0

    015

    Instruction Format

    Memory4096 X 16

    Diagram:

    Stored Program

    Organization

  • 8/7/2019 2. General System Architecture

    7/28

    Stored Program Concept (Cont)

    The Opcode tells us the operation to be performed.Address tells us the memory location where to find operand.For a memory unit of 4096 bits - 12 bits to specify address.When we store an instruction code in memory, 4 bits are specified

    for 16 operations (as 12 bits are for operand address).

    For an operation control fetches the instruction from memory, itdecodes the operation (one out of 16) and finds out the operandsand then do the operation.

    Computers with one processor register generally name itaccumulator (or AC). The operation is performed with operand andthe content of AC.

    In case no operand is specified, we compute the operation onaccumulator .E.g.: Clear AC, complement AC etc.

    MANAV

    RACHNACOLL

    EGEOFENGINE

    ERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    8/28

    Parallel Computers

    The one we studied was very basic one but sometimes wehave very large computations in which one processor withgeneral architecture will not of much help. Thus we take thehelp of many processors or divide the processor functionsinto many functional units and also doing the same

    computation on many data values.

    So to give solutions to all these we have various types ofcomputers.

    Flynns Classification

    Fengs ClassificationHndlers Classification

    MANAV

    RACHNACOLL

    EGEOFENGINE

    ERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    9/28

    Flynns Classification

    Based on the multiplicity ofInstruction Streamsand Data Streams

    Instruction Stream - Sequence of Instructionsread from memory

    Data Stream - Operations performed on the datain the processor MA

    NAV

    RACHNACOLL

    EGEOFENGINE

    ERING

    Pre ared B : Nidhi Gar

    Number of DataStreams

    SIMDSISD

    MIMDMISD

    MultipleSingle

    Single

    Multiple

    Number OfInstruction

    Streams

  • 8/7/2019 2. General System Architecture

    10/28

    Flynns Classification

    Architecture Categories

    SISD

    (Single

    Instruction

    Single

    Data)

    SIMD

    (Single

    Instruction

    Multiple

    Data)

    MISD

    (Multiple

    Instructions

    Single

    Data)

    MIMD

    (Multiple

    Instructions

    Multiple

    Data)

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    11/28

    SISD

    Simplest Computer AvailableContains No Parallelism

    Instruction are executed sequentiallySystem may or may not have externalparallel processing capabilities.

    C P MIS IS DS

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    12/28

    SISD

    CharacteristicsStandard von Neumann machineInstructions and data are stored in memory

    One operation at a timeLimitations

    Von Neumann bottleneckMaximum speed of the system is limited by the

    Memory Bandwidth (bits/sec or bytes/sec)Memory is shared by CPU and I/OExamples: Superscalar processors

    Super pipelined processorsVLIW

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    13/28

    SIMD

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

    Single instruction isoperated with differentsets of data, with the helpof many processing units.

    Controlled by a singlecontrol unit.

    Shared memory containsvarious modules so that it

    can communicate with allthe processors at sametime.

    Main memory is used for

    storage of programs.

    Control Unit

    P1

    Alignment Network

    Data bus

    Instruction Stream

    Data stream

    Processor

    units

    Memory

    P2 Pk

    M1 M2 Mk

    Memory

    Modules

  • 8/7/2019 2. General System Architecture

    14/28

    SIMD

    Characteristics

    Only one copy of the program existsA single controller executes one instruction at a

    time

    Examples:

    Array processorsSystolic arraysAssociative processors

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    15/28

    Instruction stream

    M1 C1 P1

    M2 C2 P2

    MnCn Pn

    Memory

    Data

    Stream

    MISD

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

    No Practical usage as there are least chanceswhere lots of instructions get executed on a singledata.

  • 8/7/2019 2. General System Architecture

    16/28

    MIMD

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

    Refers to a computer system where we have

    different processing elements working ondifferent data.

    In this we classify various multiprocessors andmulti computers.

    Interconnection Network

    P1 M1 Pn MnP2 M2

    Shared Memory

  • 8/7/2019 2. General System Architecture

    17/28

    MIMD

    Characteristics:

    Multiple processing units Execution of multiple instructions on multiple

    data

    Types of MIMD computer systems

    Shared memory multiprocessors1. UMA

    2. NUMA

    Message-passing multi computers

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    18/28

    Shared Memory Multi-Processors

    Example systems

    Bus and cache-based systemsSequent Balance, Encore MultimaxMultistage IN-based systemsUltra computer, Butterfly, RP3, HEPCrossbar switch-based systems

    C.mmp, Alliant FX/8Limitations

    Memory access latency Hot spot problem

    MANAV

    RACHNACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    19/28

    UMA

    All processors have equally direct access to onelarge memory address space.

    The access time to reach that memory is same forall processors thus it is named as UMA.

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

    Interconnection Network

    M1 M2 Mn

    PnP2P1

  • 8/7/2019 2. General System Architecture

    20/28

    NUMA

    All processors have equally direct access to onelarge memory address space and also have theirown memory.

    The access time to reach different memories isdifferent for each processor thus it is named as

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

    Interconnection Network

    M M M

    PnP2P1

    M1 M2 Mn

  • 8/7/2019 2. General System Architecture

    21/28

    Message Passing Multi-computer

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

    Message Passing Network

    PnP2P1

    M M M

  • 8/7/2019 2. General System Architecture

    22/28

    Message Passing Multi-computer

    CharacteristicsInterconnected computersEach processor has its own memory, andcommunicates via message-passing

    Example systems

    Tree structure: Teradata, DADOMesh-connected: Rediflow, Series 2010, J-Machine

    Hypercube: Cosmic Cube, iPSC, NCUBE, FPS TSeries, Mark IIILimitations

    Communication overhead

    Hard to programming

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    23/28

    Multilevel View-point of Machine

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEERING

    Pre ared B : Nidhi Gar

    INSTRUCTION SET ARCHITECTURE

    (ISA)

    PROCESSOR MEMORY I/0 SYSTEM

    GATE LEVEL DESIGN

    CIRCUIT LEVEL DESIGN

    SILICON LAYOUT LAYER

    COMPILER

    ASSEMBLER

    OS MSDOS

    WINDOWS

    UNIX / LINUX

    USER APPLICATION LAYERSOFTWARE

    LAYER

    HARDWARE

    LAYER

    DATA PATH & CONTROL

    MACRO

    ARCHITECTURE

    MICROARCHITECTURE

  • 8/7/2019 2. General System Architecture

    24/28

    Macro Architecture

    as a unit of deployment, talks about Clientapplications & COM Servers.

    This is basically our software layer of the computer. It comprises of :

    User Application layerHigh level languageAssembly languageMachine languageOperating systemCompiler

    Assembler

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEE

    RING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    25/28

    Operating System

    Functions of Operating system:Process managementMemory managementFile managementDevice managementError DetectionSecurity

    Types of Operating system:Multiprogramming Operating SystemMultiprocessing Operating systemTime Sharing Operating systemReal time Operating systemDistributed Operating systemNetwork Operating system

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEE

    RING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    26/28

    ISA

    Abstraction on the interface between hardware andlow-level software.

    Deals with the functional behaviour of a computersystem as viewed by a programmer.

    Attribute of a computing system, as seen byassembly language programmer or compiler.

    ISA is determined by:

    Data Storage.Memory Addressing Modes.Operations in the Instruction Set.Instruction Formats.Encoding the Instruction Set.Compilers View.

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEE

    RING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    27/28

    Micro - Architecture

    inside a unit of deployment talks about running process, threadconcurrency, synchronization, memory sharing etc.Computer organization - how constituent parts of system areinterconnected & how they interoperate in order to implement theISA.

    Processor memory I /o system Basic hardware devices requiredfor processing of any system application.Data path and control Decides the flow of information withinvarious parts of computer system in various circuits.Gate level design Circuits such as register, counters etc areimplemented in the form of various gates available.

    Circuit level design Gates forming logical circuit/componentSilicon layout layer

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEE

    RING

    Pre ared B : Nidhi Gar

  • 8/7/2019 2. General System Architecture

    28/28

    Structured Organization

    MANAV

    RACHN

    ACOLL

    EGEOFENGINEE

    RIN

    P d B Nidhi G

    Secondary Memory orAuxiliary Memory

    (Hard-disk)

    Primary Memory orMain Memory

    (RAM)

    Cache

    ALU (Data Path)

    Control Unit

    Registe

    rs

    Input

    (Instructions &Data fromkeyboard

    Or

    Other I/ODevices)

    Output

    (To monitors,Printer or otherO/P Devices)