Original Author:Adrian J Pullin Adapted by: Yanru Bao, Tianjin University

William Stallings Computer Organization and Architecture

Chapter 2Computer Evolution and Performance

2

Topics

History of Computers 计算机简史Designing for Performance 性能设计Pentium Evolution 奔腾芯片的进展Performance Measurement 性能测量方法

3

The first Generation: Vacuum Tubes 真空管

ENIAC - background

Electronic Numerical Integrator And Computer电子数字积分器和计算机

Eckert and Mauchly 埃克特和莫克利University of PennsylvaniaTrajectory tables for weapons Started 1943Finished 1946

Too late for war effort (Quiz: When did WWII end?)

Used until 1955

4

ENIAC – details (It’s BIG)

Decimal (not binary) 采用十进制表示 / 运算 20 accumulators of 10 digits 20 个 10 位累加器 Programmed manually by switches 设置开关手工编

程 18,000 vacuum tubes 真空管 70,000 resistors 电阻 10,000 capacitors 电容 6,000 switches 开关 30 tons 15,000 square feet 140 KW power consumption 5,000 additions per second

5

6

7

8

von Neumann/Turing 冯诺依曼 / 图灵Stored Program concept

Main memory storing programs and data主存储器：用于存储数据和指令

ALU operating on binary data能够操作二进制数的算术逻辑单元

Control unit interpreting instructions from memory and executing控制器：翻译内存中的指令并执行

Input and output equipment operated by control unit由控制器操纵的输入、输出设备

9

von Neumann/Turing

存储程序思想：将事先编好的程序和原始数据送入主存中，然后启动计算机工作。计算机应能在不需操作人员干预下，自动完成逐条取出指令和执行指令的任务。

10

冯‧诺依曼生平简介

美籍匈牙利数学家 美籍匈牙利数学家

11

12

Princeton Institute for Advanced Studies 普林斯顿高级研究院 , 1946 IAS

Completed 1952:Basis for virtually every machine designed since then 是后来通用计算机的原型

13

Structure of von Nuemann machine

MainMemory

Arithmetic and Logic Unit

Program Control Unit

InputOutputEquipment

14

IAS details --- memory

1000 x 40 bit words （ figure 2.2 ） Binary number 2 x 20 bit instructions

15

IAS – von Neumann Architecture

Features Data and instructions are stored in a

single R/W memory 数据和指令使用同一个读写存储器

Memory contents are addressable by location, regardless of the content存储器内容根据地址访问

Sequential execution顺序执行

16

IAS – von Neumann Architecture

Set of registers (storage in CPU) Memory Buffer Register(MBR) 存储缓冲寄存

器 Memory Address Register(MAR) 存储地址寄

存器 Instruction Register(IR) 指令寄存器 Instruction Buffer Register(IBR) 指令缓冲寄

存器 Program Counter(PC) 程序计数器 Accumulator(AC) 累加器 Multiplier Quotient 乘商寄存器 (MQ)

17

18

IAS instruction cycle ---two subcycle

First cycle—fetch cycle 取指令周期The opcode of the next instruction is

loaded into the IR and the address portion is loaded into the MAR. This instruction may be taken from the IBR, or it can be obtained from memory by loading a word into the MBR, and then down to the IBR, IR, and MAR.

下一条指令的操作码装入 IR ，地址部分装入MAR 。指令可以从 IBR 获得。也可以通过由存储器向 MBR 装入一个字，再将它分解后放入IBR 、 IR 和 MAR 来获得。

19

IAS instruction cycle ---two cycle

Second cycle—execute cycleControl circuitry interprets the

opcode and executes the instruction by sending out the appropriate control signals to cause data to be moved or an operation to be performed by the ALU.

控制电路翻译操作码，并且通过发送相应的控制信号来执行指令。这些信号控制数据的传送和ALU 操作的执行。

20

21

冯•诺依曼结构的主要思想

22

Commercial Computers 商用计算机

1947 - Eckert-Mauchly Computer Corporation 埃克特－莫克利计算机公司

UNIVAC I (Universal Automatic Computer)通用自动计算机 , the first successful commercial computer. 第一台成功的商用计算机

US Bureau of Census 1950 calculations 由统计局委托制造，用于 1950 年计算

Became part of Sperry-Rand Corporation成为 Sperry-Rand 公司的一部分

Late 1950s - UNIVAC II Faster 更快 More memory 存储容量更大

23

24

25

IBM

Punched-card processing equipment穿孔卡片处理设备

1953 - the 701 IBM’s first stored program computer

该公司第一台存储程序计算机 Scientific calculations 科学计算

1955 - the 702 Business applications 商业应用

Lead to 700/7000 series

26

The second Generation ： Transistors晶体管

Replaced vacuum tubes取代真空管

Characteristic: Smaller 体积小 Cheaper 便宜 Less heat dissipation 散热少 Solid State device 固体元件 Made from Silicon (Sand) 原料是硅

Invented 1947 at Bell Labs1947 年由贝尔实验室发明

27

Transistor Based Computers

Second generation machines第二代机器

NCR & RCA produced small transistor machinesNCR&RCA 公司生产的小型晶体管机器

IBM 7000DEC - 1957

Produced PDP-1

28

Transistor Based Computers

The second generation saw the introduction 第二代计算机所引进的新技术 : High level languages 高级语言 Floating point arithmetic 浮点算术运算 System software 系统软件

29

The Third Generation: Integrated Circuits 集成电路 Discrete Component 分立元件

A single, self-contained transistor单个独立封装的晶体管

Electronic equipment was composed largely of discrete components （ throughout the 50s and early 60s) 20世纪 50 年代和 60 年代早期，电子设备主要由分立元件组成 transistors, resistors, capacitors

These facts of life were beginning to create problems in the computer industry.这种情况给计算机工业带来了问题

30

Microelectronics 微电子

In 1958 the era of microelectronics started. 1958 年微电子时代开始

Literally - “small electronics” “微小的电子技术” A computer is made up of gates, memory cells

and interconnections.The gates and memory cells are constructed of simple digital electronic components.计算机包含门、内存位元和它们之间的互连结构。而这些门和存储器位元是由简单的数字电子元件组成。

These can be manufactured on a semiconductor这些电子元件可以用半导体制成。

e.g. silicon wafer (see figure 2.7)

31

8 寸硅晶片，每一个小方格经过切割，封装后就是一片芯片

32

Moore’s Law摩尔定律 About increased density of

components on chip关于芯片上集成元件密度的增长速度

In 1965, Gordon Moore - cofounder of Intel 公司创始人

Number of transistors on a chip will double every year芯片上的晶体管数量每年翻一番

Since 1970’s development has slowed a little 七十年代以后发展速度减慢 Number of transistors doubles every 18

months 晶体管数量每 18个月翻一番

33

Growth in CPU Transistor Count

34

Moore’s Law摩尔定律 Profound consequences 影响深远 :

Cost of a chip has remained almost unchanged单个芯片的成本没有变化

Higher packing density means shorter electrical paths, giving higher performance集成度更高意味着电路长度更短，所以提高了性能

Smaller size gives increased flexibility计算机更小，容易放置在各种环境中

Reduced power and cooling requirements减少了电能消耗及冷却的要求

Fewer interconnections increases reliability集成电路内部的连接比焊接连接更可靠

35

IBM 360 series

1964, IBM announced the System/360. Replaced (& not compatible with) 7000 series First planned “family” of computers

Similar or identical instruction sets相同或相似的指令集

Similar or identical O/S 相似或相同的操作系统

Increasing speed 更高的速度

Increasing number of I/O ports (i.e. more terminals)更多的 I/O端口数

Increased memory size 更大的内存容量

Increased cost 成本增加

36

DEC PDP-8

1964 First minicomputer ( 小型机 ) Did not need air conditioned room

不需要空调房间 Small enough to sit on a lab bench

小到可以放在实验室的长椅上 $16,000

$100k+ for IBM 360 Embedded applications &OEM被原始设备制造商集成到自己系统中

BUS STRUCTURE总线结构

37

38

DEC - PDP-8 Bus Structure

OMNIBUS

ConsoleController

CPU Main Memory I/OModule

I/OModule

39

Later Generations:Semiconductor Memory 半导体存储器

1970Fairchild (仙童 )Size of a single core 体积是一个磁核的大小

i.e. 1 bit of magnetic core storageHolds 256 bits 可存储 256 位Non-destructive read 非破坏性读Much faster than core 比磁核速度快Capacity approximately doubles each

year存储容量每年翻一番

40

41

42

Microprocessors--Intel

1968 年，诺依斯、摩尔和格罗夫，毅然退出“仙童”，根据摩尔的建议，把新建公司取名为“英特尔”（英文缩写 INTEL ），寓意着“集成电子”。

43

Microprocessors--Intel

1971 - 4004 First microprocessor 第一个微处理器 All CPU components on a single chip

所有 CPU组件在一个芯片上 4 bit

Followed in 1972 by 8008 8 bit Both designed for specific applications

1974 - 8080 Intel’s first general purpose microprocessor

第一个通用微处理器 End of 70s, 16-bit general-purpose, 8086 1985, 32-bit, 80386

44

Generations of Computer

First generation: Vacuum tube - 1946-1957Second generation: Transistor - 1958-1964Third generation: Integrated circuits – 1965 –

1971 Small scale integration - 1965 on

Up to 100 devices on a chip Medium scale integration - to 1971

100-3,000 devices on a chip Semiconductor memory (1970) Microprocessor (1971)

45

Generations of Computer

Fourth generation: Large scale integration (LSI) - 1971-1977 3,000 - 100,000 devices on a chip Intel 8080: first general-purpose

microprocessor (1974)Fifth generation: 1978 – present

Very large scale integration (VLSI 超大规模 ) - 1978 to date

100,000 - 100,000,000 devices on a chip Ultra large scale integration (ULSI 甚大规模 )

Over 100,000,000 devices on a chip Giga scale integration (GSI) – 十亿，千兆

46

Designing for Performance 1

Support-Demand CycleComputer Performance 计算机性能

Demands Supports (Motivates)Application Requirement 应用需求

47

Designing for Performance 2

Performance balance—computer design is a constantly evolving art form, why? 计算机设计是一个不断进化的艺术形式 The rate at which performance is changing in

the various technology areas (processor, buses, memory, peripherals) differs greatly from one type of elements to another.从一种类型的元器件到另一种类型的元器件，对于各种不同的技术领域，它们性能提高的速率相差很大

New applications and new peripheral devices constantly change the nature of the demand on the system.新的应用和新的外围设备不断改变对系统特性的要求

48

Speeding it up (Processor)

Branch prediction:The processor looks ahead in the software and predicts which branches are likely to be processed next.

转移预测：处理器在软件中趋向于顺序向前执行，并可预测哪条分支指令集或哪组指令将会被执行

49

Speeding it up (Processor)

Data flow analysis:The processor analyzes which instructions are dependent on each other’s results,or data, to create an optimized schedule of instructions.

数据流分析：处理器分析哪一条指令依赖于其他的结果或数据，来优化指令调度。

50

Speeding it up (Processor)

Speculative execution:Using previous two methods,some processors speculatively execute instructions ahead of their actual appearance, holding the results in temporary locations.

推测执行：使用转移预测和数据流分析，一些处理器让指令在程序实际执行之前就“推测执行”，并把结果暂时存储起来。通过执行可能需要的指令，使处理器的执行机制尽可能的保持繁忙。

51

Performance Mismatch 性能不匹配

Processor speed increased处理器速度加快

Memory capacity increased存储器容量增加

Memory speed lags behind processor speed但存储器发展的速度落后于处理器发展

52

DRAM and Processor Characteristics

53

Trends in DRAM use

54

Solutions(1-2)

Increase number of bits retrieved at one time Make DRAM “wider” rather than “deeper”

and By using wide bus data paths使 DRAM 的接口“更宽”而不是“更深”，即增大总线的

数据宽度，来增加每次所能取出的位数Change DRAM interface to make it more

efficient Including a cache on the DRAM chip通过在 DRAM 芯片中加入高速缓存或其他缓冲机制来改进

DRAM接口，提高它的效率

55

Solutions(3-4)

Reduce frequency of memory access More complex cache between memory and

proccesor通过在主存和处理器之间引入更复杂、更有效的高速缓存结

构，来减少存储器访问频度。 Increase interconnection bandwidth

High speed buses Hierarchy of buses通过高速总线和使用分层总线来缓冲和结构化数据流。

56

Pentium Evolution (1)

8080 first general purpose microprocessor 8 bit data path Used in first personal computer – Altair

8086 much more powerful 16 bit instruction cache, prefetch few instructions 8088 used in first IBM PC

57

Pentium Evolution (2)

80286 16 Mbyte memory addressable

80386 32 bit Support for multitasking

80486 sophisticated powerful cache and

instruction pipelining built in maths co-processor

58

Pentium Evolution (3)

Pentium Superscalar （超标量） :Multiple

instructions executed in parallel

Pentium Pro Increased superscalar organization register renaming branch prediction data flow analysis speculative execution （推测执行）

59

Pentium Evolution (4)

Pentium II MMX technology: graphics, video & audio processing

Pentium III Additional floating point instructions for 3D graphics

Pentium 4 Note Arabic rather than Roman numerals Further floating point and multimedia enhancements

Itanium安腾 64 bit workstation processor

See www.intel.com for detailed information on processors