Transcript

1

Computer Organization and Architecture

IntroductionChapters 1-2

Architecture & Organization 1

zArchitecture is those attributes visible to the programmeryInstruction set, number of bits used for data

representation, I/O mechanisms, addressing techniques.ye.g. Is there a multiply instruction?

zOrganization is how features are implemented, typically hidden from the programmeryControl signals, interfaces, memory technology.ye.g. Is there a hardware multiply unit or is it done by

repeated addition?

2

Architecture & Organization 2

z All Intel x86 family share the same basic architecturez The IBM System/370 family share the same basic

architecture

z This gives code compatibilityyAt least backwardsyBut… increases complexity of each new generation. May be

more efficient to start over with a new technology, e.g. RISC vs. CISC

z Organization differs between different versions

Structure & Function

zComputers are complex; easier to understand if broken up into hierarchical components. At each level the designer should consideryStructure : the way in which components relate to

each otheryFunction : the operation of individual components

as part of the structure

zLet’s look at the computer top-down starting with function.

3

Function

zAll computer functions are:yData processingyData storageyData movementyControl

Functional view

zFunctional view of a computer

DataMovementApparatus

ControlMechanism

DataStorageFacility

DataProcessingFacility

4

Operations (1)

zData movementye.g. keyboard to screen

DataMovementApparatus

ControlMechanism

DataStorageFacility

DataProcessingFacility

Operations (2)

zStorage ye.g. Internet download to disk

DataMovementApparatus

ControlMechanism

DataStorageFacility

DataProcessingFacility

5

Operation (3)

zProcessing from/to storage ye.g. updating bank statement

DataMovementApparatus

ControlMechanism

DataStorageFacility

DataProcessingFacility

Operation (4)

zProcessing from storage to I/Oye.g. printing a bank statement

DataMovementApparatus

ControlMechanism

DataStorageFacility

DataProcessingFacility

6

Structure

zMajor Components of a ComputeryCentral Processing Unit (CPU) – Controls the

operation of the computer and performs data processingyMain Memory – Stores datayInput Output (I/O) – Moves data between the

computer and the external environmentySystem Interconnect – Some mechanism that

provides for communications between the system components, typically a bus (set of wires)

Structure - Top Level

Computer

Main Memory

InputOutput

SystemsInterconnection

Peripherals

Communicationlines

CentralProcessing Unit

Computer

7

Structure - CPU

zMajor components of the CPUyControl Unit (CU) – Controls the operation of the CPUyArithmetic and Logic Unit (ALU) – Performs data

processing functions, e.g. arithmetic operationsyRegisters – Fast storage internal to the CPU, but

contents can be copied to/from main memoryyCPU Interconnect – Some mechanism that provides

for communication among the control unit, ALU, and registers

Structure - The CPU

Computer Arithmeticand Login Unit

ControlUnit

Internal CPUInterconnection

Registers

CPU

I/O

Memory

SystemBus

CPU

8

Structure – Inside the CPU

zThe implementation of registers and the ALU we will leave primarily to EE 241zWe will say a bit about the architecture of the

control unit, there are many possible approaches. yA common approach is the microprogrammed control

unit, where the control unit is in essence itself a miniature computer, where a CPU instruction is implemented via one or more “micro instructions”⌧Sequencing Logic – Controlling the order of events⌧Microprogram Control Unit – Internal controls

⌧Microprogram Registers, Memory

Structure – A MicroprogrammedControl Unit

CPU

ControlMemory

Control Unit Registers and Decoders

SequencingLogin

ControlUnit

ALU

Registers

InternalBus

Control Unit

9

Outline of the Stallings Text (1)

zComputer Evolution and PerformancezComputer Interconnection StructureszInternal MemoryzExternal MemoryzInput/OutputzOperating Systems SupportzComputer ArithmeticzInstruction Sets

Outline of the Stallings Text (2)

zCPU Structure and FunctionzReduced Instruction Set ComputerszSuperscalar ProcessorszControl Unit OperationzMicroprogrammed ControlzMultiprocessors and Vector ProcessingzDigital Logic (Appendix)

10

Computer Evolution and Performance

Better, Faster, Cheaper?

History: ENIAC background

zElectronic Numerical Integrator And ComputerzEckert and MauchlyzUniversity of PennsylvaniazTrajectory tables for weapons, BRLzStarted 1943zFinished 1946yToo late for war effort

zUsed until 1955

11

ENIAC - details

zDecimal (not binary)z20 accumulators of 10 digits (ring of 10 tubes)zProgrammed manually by switchesz18,000 vacuum tubesz30 tonsz15,000 square feetz140 kW power consumption (about $10/hr today)z5,000 additions per second

Vacuum Tubes

Grid regulates flow from of electrons from the cathode

12

von Neumann/Turing

zENIAC : Very tedious to manually wire programszvon Neumann architecture:yStored Program conceptyMain memory storing programs and datayALU operating on binary datayControl unit interpreting instructions from memory and

executingyInput and output equipment operated by control unityPrinceton Institute for Advanced Studies ⌧IAS

yCompleted 1952

13

Structure of von Neumann machine

MainMemory

Arithmetic and Logic Unit

Program Control Unit

InputOutputEquipment

IAS - detailsz 1000 x 40 bit wordsyBinary numbery2 x 20 bit instructions

z Set of registers (storage in CPU)yMemory Buffer RegisteryMemory Address RegisteryInstruction RegisteryInstruction Buffer RegisteryProgram CounteryAccumulatoryMultiplier Quotient

0 1 39

Sign bit

Number Word

Instruction Word

0 8 20 28 39

LeftOpCode Address

RightOpCode Address

14

Structure of IAS - detail

MainMemory

Arithmetic and Logic Unit

Program Control Unit

InputOutputEquipment

MBR

Arithmetic & Logic Circuits

MQAccumulator

MAR

ControlCircuits

IBR

IR

PC

Address

Instructions& Data

Central Processing Unit

IAS Instruction Cycle

z The IAS repetitively performs the instruction cycle:yFetch⌧Opcode of the next instruction is loaded into the IR⌧Address portion is loaded into the MAR⌧Instruction either taken from the IBR or obtained from memory by

loading the PC into the MAR, memory to the MBR, then the MBR to the IBR and the IR

• To simplify electronics, only one data path from MBR to IR

yExecute⌧Circuitry interprets the opcode and executes the instruction⌧Moving data, performing an operation in the ALU, etc.

z IAS had 21 instructionsyData transfer, Unconditional branch, conditional branch,

arithmetic, address modification

15

Commercial Computers

z1947 - Eckert-Mauchly Computer CorporationzUNIVAC I (Universal Automatic Computer)zUS Bureau of Census 1950 calculationszBecame part of Sperry-Rand CorporationzLate 1950s - UNIVAC IIyFasteryMore memoryyUpward compatible with older machines

IBM

zPunched-card processing equipmentz1953 - the 701yIBM’s first stored program computeryScientific calculations

z1955 - the 702yBusiness applications

zLead to 700/7000 series

16

Transistors

zReplaced vacuum tubeszSmallerzCheaperzLess heat dissipationzSolid State devicezMade from Silicon (Sand)zInvented 1947 at Bell LabszShockley, Brittain, Bardeen

Transistor Based Computers

zSecond generation of machineszNCR & RCA produced small transistor machineszIBM 7000zDEC - 1957yProduced PDP-1

17

IBM 7094

zLast member of the 7000 seriesy50 times faster than the 701⌧1.4 uS vs. 30 uS cycle

y32K memory vs. 2KyMain memory: Core memory vs. TubesyCPU memory: transistors vs. Tubesy185 vs. 24 opcodesyInstruction fetch overlap, reduced another trip to

memory (exception are branches)yData channels, independent I/O module for devices

3rd Generation: Integrated Circuits

zSelf-contained transistor is a discrete componentyBig, manufactured separately, expensive, hot when

you have thousands of them

zIntegrated CircuitsyTransistors “etched” into a substrate, bundled

together instead of discrete componentsyAllowed thousands of transistors to be packaged

together efficiently

18

Microelectronics

zLiterally - “small electronics”zA computer is made up of gates, memory cells

and interconnectionszThese can be manufactured on a

semiconductor, e.g. silicon waferyThin wafer divided into chipsyEach chip consists of many gates/memory cellsyChip packaged together with pins, assembled on a

printed circuit board

Generations of Computer

z Vacuum tube - 1946-1957z Transistor - 1958-1964z Small scale integration - 1965 onyUp to 100 devices on a chip

z Medium scale integration - to 1971y100-3,000 devices on a chip

z Large scale integration - 1971-1977y3,000 - 100,000 devices on a chip

z Very large scale integration - 1978 to datey100,000 - 100,000,000 devices on a chipyPentium IV has about 40 million transistors

z Ultra large scale integrationyOver 100,000,000 devices on a chip (vague term)

19

Moore’s Lawz Increased density of components on chipz Gordon Moore : co-founder of Intelz Number of transistors on a chip will double every yearz Since 1970’s development has slowed a littleyNumber of transistors doubles every 18 months

z Cost of a chip has remained almost unchangedz Higher packing density means shorter electrical paths,

giving higher performancez Smaller size gives increased flexibilityz Reduced power and cooling requirementsz Fewer interconnections increases reliability

Growth in CPU Transistor Count

20

IBM 360 series

z 1964z Replaced (& not compatible with) 7000 seriesyReason: Needed to break out of constraints of the 7000

architecture

z First planned “family” of computersySimilar or identical instruction setsySimilar or identical O/SyIncreasing speedyIncreasing number of I/O ports (i.e. more terminals)yIncreased memory size yIncreased cost (not always the case today!)

z Multiplexed switch structure

DEC PDP-8

z1964zFirst minicomputer (after miniskirt!)zDid not need air conditioned roomzSmall enough to sit on a lab benchz$16,000 y$100k+ for IBM 360

zEmbedded applications & OEMzBUS STRUCTURE

21

DEC - PDP-8 Bus Structure

OMNIBUS

ConsoleController

CPU Main Memory I/OModule

I/OModule

96 separate signal paths to carry control, address, data signalsHighly flexible, allowed modules to be plugged in for different configurations

Other Innovations -Semiconductor Memory

z1970zFairchildzSize of a single coreyi.e. 1 bit of magnetic core storageyHeld 256 bits

zNon-destructive readzMuch faster than corezCapacity approximately doubles each year

22

Intel

z 1971 - 4004 yFirst microprocessoryAll CPU components on a single chipy4 bit

z Followed in 1972 by 8008y8 bityBoth designed for specific applications

z 1974 - 8080yIntel’s first general purpose microprocessor

z Evolution: 8086, 8088, 80286, 80386, 80486, Pentium Pentium Pro, Pentium II, Pentium III, Pentium IV, Itanium

Speeding it up

zSmaller manufacturing process (0.11 micron)zPipeliningzOn board cachezOn board L1 & L2 cachezBranch predictionzData flow analysiszSpeculative executionzParallel execution

23

Performance Mismatch

zProcessor speed increasedzMemory capacity increasedzMemory speed lags behind processor speed

zCommon memory chip technologyyDRAM = Dynamic Random Access Memory

DRAM and Processor Characteristics

24

Trends in DRAM use

Solutions

z Increase number of bits retrieved at one timeyMake DRAM “wider” rather than “deeper”

z Change DRAM interfaceyCache

z Reduce frequency of memory accessyMore complex cache and cache on chip

z Increase interconnection bandwidthyHigh speed busesyHierarchy of buses

z Similar problems with I/O devices, e.g. graphics, networkz Need balance in computer design


Top Related