Transcript CSI/CCE Computer Architecture

Introduction to Computer Architecture
Lecture 1 – Introduction
August 18th, 2008
www.qatar.cmu.edu
Computer Architecture
Fall 2008 ©
Teaching Staff
• Instructors
- Prof. Majd F. Sakr ([email protected])
- Prof. Nael Abu-Ghazaleh ([email protected])
• TA
- Adnan Majeed ([email protected])
Computer Architecture
Fall 2008 ©
Where Do We Find a Computer/Processor?
Planes
ATMs
ipod
PDA
Cameras
Cars
Watch
Cell phones
Traffic
Controller
Design &
Engineering
Music
Robots
Games
Microwave
Medical
(MRI)
Computer Architecture
Fall 2008 ©
Why Did We Develop Computers?
A solution to a problem!
While thinking of a solution,
think about:
Problem
Solution
• Cost $$$
Implementation
• Speed
Computer
• Energy/Power
Result
• Size
• Efficiency
• etc…
Computer Architecture
Fall 2008 ©
Types of Computers
°Personal Computer
°Workstation
°Server
°Supercomputer
°Embedded
Computer Architecture
Fall 2008 ©
Number of Computers Sold
Embedded
Desktops
Servers
Millions of Computers
1200
1000
1122
892
862
800
600
400
200
488
290
135
114
93
4
3
3
129
4
131
5
0
1998
1999
2000
Computer Architecture
2001
2002
Fall 2008 ©
Computer Architecture
Our Area of Understanding
Problem
Solution
Our Area of
Focus
Implementation
Compiler
Computer
Result
Computer Architecture
Fall 2008 ©
Where is “Computer Architecture and Engineering”?
Application (MediaPlayer)
Compiler
Software
Hardware
Assembler
Operating
System
(Windows XP)
Processor Memory I/O system
Instruction Set
Architecture
Datapath & Control
Digital Design
Circuit Design
Architecture
transistors
* Coordination of many levels of abstraction
Computer Architecture
Fall 2008 ©
Anatomy: 5 components of any Computer
Personal Computer
Computer
Processor
Control
(“brain”)
Datapath
(“work”)
Memory
Devices
(where
programs
&
data
live when
running)
Input
Output
Keyboard,
Mouse
Disk
(where
programs
& data
live when
not running)
Display,
Printer
Computer Architecture
Fall 2008 ©
Computer Technology - Dramatic Change!
°Processor
• 2X in speed every 1.5 years (since ‘85);
100X performance increase in last decade.
°Memory
• DRAM capacity: 2x / 2 years (since ‘96);
64x size improvement in last decade.
°Disk
• Capacity: 2X / 1 year (since ‘97)
• 250X size increase in last decade.
Computer Architecture
Fall 2008 ©
Tech. Trends: Microprocessor Complexity
2 * transistors/Chip Every 1.5 to 2.0 years
Called “Moore’s Law”
Computer Architecture
Fall 2008 ©
Architecture & Organization
°Computer Architecture
• What the “low level” programmer sees
- Types of Instructions
- Number of Registers
- Types of Operations
°Computer Organization
• How the designer Implements the Design
- Layout
- Interconnection (wires)
Computer Architecture
Fall 2008 ©
Computer Architecture and Organization
Application (MediaPlayer)
Compiler
Software
Assembler
Operating
System
(Windows XP)
Instruction Set Architecture
Hardware
Processor Memory I/O system
Architecture
Datapath & Control
Layout & Technology
Digital Design
Circuit Design
Organization
Transistors
Computer Architecture
Fall 2008 ©
Architecture & Organization 1
° Architecture is those attributes visible to the
programmer
• Instruction set, number of bits used for data representation,
I/O mechanisms, addressing techniques.
• e.g. Is there a multiply instruction?
° Organization is how features are implemented
• Control signals, interfaces, memory technology.
• e.g. Is there a hardware multiply unit or is it done by repeated
addition?
Computer Architecture
Fall 2008 ©
Architecture & Organization 2
°All Intel x86 family share the same basic
architecture
°The IBM System/370 family share the same
basic architecture
°This gives code compatibility
• At least backwards
°Organization might highly differ between
different versions
Computer Architecture
Fall 2008 ©
Course Path
Input
Multiplier
Input
Multiplicand
Instruction Sets
32
Multiplicand
Register
LoadMp
32=>34
signEx
<<1
32
34
34
32=>34
signEx
opcode
rs
rt
rs
rt
rd
shamt
funct
1
0
34x2 MUX
Arithmetic
Multi x2/x1
34
34
Sub/Add
34-bit ALU
Control
Logic
funct
32
opcode
rs
rt
immediate
opcode
rs
rt
32
rd
shamt
Result[HI]
CPU
“Moore’s Law”
100
ENC[2]
ENC[1]
ENC[0]
LO[1:0]
32
Result[LO]
funct
Performance
00
LO register
(16x2 bits)
2
LoadHI
offset
2
Prev
rt
ShiftAll
HI register
(16x2 bits)
Booth
Encoder
rs
2
LO[1]
Extra
2 bits
opcode
32
2
LoadLO
shamt
ClearHI
rd
opcode
"LO
[0]"
34
µProc
60%/yr.
(2X/1.5yr)
Processor-Memory
Performance Gap:
(grows 50% / year)
10
DRAM
9%/yr.
DRAM (2X/10 yrs)
Computer Architecture
Fall ‘08
Time
9
200
0
1
199
2
199
3
199
4
199
5
199
6
199
7
199
8
199
6
198
7
198
8
198
9
199
0
199
198
0
198
1
198
2
198
3
198
4
198
5
198
1
Datapaths &
Control
C
I/O P
U
Memory Systems
Computer Architecture
Y
O
U
R
Fall 2008 ©
Homeworks and Projects
°Quizzes (weekly)
°Assignment (every ~2 weeks)
°Project (every ~3-4 weeks)
°End of Semester Project:
• Demo
• Oral Presentation
• Head-to-head Race
• Final Report
Computer Architecture
Fall 2008 ©
Course Exams
°Reduce the pressure of taking exams
• Exam I
• Exam II
• Final
°Goal
• Our goal: test knowledge vs. speed writing
(no memorization)
• Review meetings: before?
Computer Architecture
Fall 2008 ©
Grading
°Grade breakdown
• Exam I:
• Exam II:
• Final:
• Projects
• Homeworks
• Quizzes
• Attendance/Participation:
10%
10%
20%
40%
10%
5%
5%
°No late homeworks or projects!
°Written request for changes to grades
Computer Architecture
Fall 2008 ©
Our Goals
° Show you how to understand modern computer architecture
in its rapidly changing form
° Show you how to design by leading you through the process
on challenging design problems and by examining real
designs
° Learn application analysis and new design techniques
Computer Architecture
Fall 2008 ©
Text
° Required: Computer Organization and
Design,
3rd Edition, Patterson and Hennessy
(COD)
° Reference: Computer Organization
and Architecture, 6thEdition,
William Stallings
• Readings on web page
http://williamstallings.com/COA6e.html
° Reference: Structured Computer Organization,
4th Edition, Andrew S. Tanenbaum
Computer Architecture
Fall 2008 ©
The
Big Picture
Computer Architecture
Fall 2008 ©
Types of Processors
Computer Architecture
Fall 2008 ©
Hardware/Software Divide
Excel
Internet Explorer
Visual Studio
Application
Hardware
Windows XP
Linux
Solaris
OS X
PC
MAC
SUN
Computer Architecture
Fall 2008 ©
Program Path to Execution
High Level
Language Program
(.c file)
Compiler
Assembly Language
Program (.asm file)
Assembler
Binary Machine
Language Program
(.exe file)
Computer Architecture
Fall 2008 ©
The Five Components of a Computer
Computer Architecture
Fall 2008 ©
The Motherboard:
The five von Neumann
components:
Input
&
Output
ALU
&
CU
M
Computer Architecture
Fall 2008 ©
Motherboard
Computer Architecture
Fall 2008 ©
Inside the Processor
Computer Architecture
Fall 2008 ©
Manufacturing Process
Computer Architecture
Fall 2008 ©
An 8-inch (200-mm) Diameter Wafer
Computer Architecture
Fall 2008 ©
Modern Fabs
°Current minimum feature size is 45nano
meters (45x10-9 meters)
°Can fit over a million transistors on the tip
of a hair
°Fab facility costs 3 billion US $
• Many chip designers are fab-less
°Employs 100s of employees
°Yield on the order of 30%
Computer Architecture
Fall 2008 ©
Computer’s History
1st generation: Vacuum Tubes
°During World War 2 the
Army’s Ballistics Research
Laboratory employed more
than 200 people to solve
essential ballistics equations
using desktop calculators.
Computer Architecture
Fall 2008 ©
1st generation: Vacuum Tubes
Professor Mauchly (EE)
& his gradate student Eckert
proposed to build a general purpose
computer using vacuum tubes for the
Ballistics Research Laboratory (BRL)
Computer Architecture
Fall 2008 ©
ENIAC (Electronic Numerical Integrator And Computer)
° ENIAC built in World War II was the first general purpose computer
• Used for computing artillery firing tables
• 24 meters long by 2.5 meters high and several meters wide
• Each of the twenty 10 digit registers was 1 meter long
–Since then:
Moore’s Law:
transistor capacity doubles
every 18-24 months
Computer Architecture
Fall 2008 ©
1st generation: ENIAC Completed in 1946
Programming the
ENIAC
° Decimal (not binary)
° 20 accumulators of 10 digits
° Programmed manually by switches &
cables
0 1 2
° 18,000 vacuum tubes
3
9
4
8
° 30 tons
5
7
° 15,000 square feet
° 140 kW power consumption
° 5,000 additions per second
Computer Architecture
Fall 2008 ©
6
The von Neuman machine - Completed 1952
° Stored Program concept
Scientist at the
Institute of
Advanced
Studies
° Main memory storing programs and
data
° ALU operating on binary data
° Control unit interpreting instructions
from memory and executing
° Input and Output equipment operated
by control unit
Computer Architecture
Fall 2008 ©
Structure of von Neumann Machine
Central Processing Unit
CPU
CA
Arithmetic –Logic Unit
Main
Memory
M
Input/Output
Equipment
Program Control Unit
CC
I/O
R
Computer Architecture
Fall 2008 ©
Commercial Computers
° 1947 - Eckert-Mauchly Computer Corporation
° 1st successful machine:
UNIVAC I (Universal Automatic Computer)
° Commissioned by the US Bureau of Census for
the 1950 calculations
° Became part of Sperry-Rand Corporation
° Late 1950s - UNIVAC II
• Faster
• More memory
• Upward Compatibility
Computer Architecture
Fall 2008 ©
2nd Generation: Transistors
°Replaced vacuum tubes
°Smaller & Cheaper
°Less heat dissipation
°Solid State device (silicon)
°Invented 1947 at Bell Labs
The First Transistor
Computer Architecture
Fall 2008 ©
Transistor Based Computers
°Second generation machines
°NCR & RCA produced small transistor
machines
°IBM 7000
°DEC - 1957
• Produced PDP-1
Computer Architecture
Fall 2008 ©
Microelectronics
°Literally - “small electronics”
°A computer is made up of gates, memory
cells and interconnections
°These can be manufactured on a
semiconductor
°e.g. silicon wafer
Computer Architecture
Fall 2008 ©
Growth in CPU Transistor Count
Computer Architecture
Fall 2008 ©
Moore’s Law
° Increased density of components on chip
° Gordon Moore - cofounder of Intel
° Number of transistors on a chip will double
every year
° Since 1970’s development has slowed a little
• Number of transistors doubles every 18
months
° Cost of a chip has remained almost unchanged
Computer Architecture
Fall 2008 ©
Moore’s Law - Cont’d
° Higher packing density means shorter
electrical paths, giving higher performance
° Smaller size gives increased flexibility
° Reduced power and cooling requirements
° Fewer interconnections increases reliability
Computer Architecture
Fall 2008 ©
Moore’s Law—Will it continue?
°A number of “walls” on the horizon
• Physical process wall: impossible to
continue shrinking transistor sizes
- Already leading to low yield, soft-errors, process
variations
• Power wall
- Power consumption and density have also been
increasing
• Other issues:
-
What to do with the transistors?
Wire delays
Memory and I/O walls
New architectures? Multi-cores
Computer Architecture
Fall 2008 ©
Yield Trends with Process Size
Computer Architecture
Fall 2008 ©
Computer Architecture
Fall 2008 ©
Computer Architecture
Fall 2008 ©
Computer Generations
Generation
Dates
Technology
1
1946-1957
Vacuum Tube
Operations per
Second
40,000
2
1958-1964
Transistor
200,000
3
1965-1971
1,000,000
4
1972-1977
5
1978-…
Small & Medium
Scale Integration
Large Scale
Integration (LSI)
Very Large Scale
Integration (VLSI)
Computer Architecture
10,000,000
100,000,000
Fall 2008 ©
And in conclusion...
°Continued rapid improvement in Computing
• 2X every 1.5 years in processor speed;
every 2.0 years in memory size;
every 1.0 year in disk capacity;
Moore’s Law enables processor, memory
(2X transistors/chip/ ~1.5 ro 2.0 yrs)
°5 classic components of all computers
Control Datapath Memory Input Output
Processor
Computer Architecture
Fall 2008 ©