Datorarkitektur och operativsystem
U
Unmesh
h D.
D Bordoloi
B d l i
The first electronic computers
ENIAC
 J. Presper Eckert and John Mauchly at the Moore
School of the University of Pennsylvania
 Funded by the United States Army
 B
Became operational
ti l d
during
i W
World
ld W
War II but
b t was nott
publicly disclosed until 1946
ENIAC (Electronic Numerical Integrator and Calculator) –
The world’s first electronic computer
ENIAC
Von Neumann
 ENIAC pro vided
id d conditional
di i l jumps
j
and
d was
programmable, clearly distinguishing it from earlier
calculators
 In 1944
1944, JJohn
hn von
n Ne
Neumann
mann was
as attracted to
t the ENIAC project.
r ject The
group wanted to improve the way programs were entered and discussed
storing programs as numbers; von Neumann helped crystallize the ideas
p
g a stored-program
p g
computer
p
called EDVAC
and wrote a memo pproposing
(Electronic Discrete Variable Automatic Computer).
 Programming was done manually by plugging cables and
setting switches, and data was entered on punched cards.
Programming
g
g for typical
yp
calculations required
q
from half
an hour to a whole day
 Herman Goldstine distributed the memo and pput von Neumann’s name on
it, much to the dismay of Eckert and Mauchly, whose names were omitted.
This memo has served as the basis for the commonly used term von
Neumann computer. Several early pioneers in the computer field believe
that this term gives too much credit to von Neumann,
Neumann who wrote up the
ideas, and too little to the engineers, Eckert and Mauchly, who worked on
the machines.
 ENIAC was a general-purpose machine, limited primarily
by a small amount of storage and tedious programming
 For this reason, the term does not appear elsewhere in your textbook
book.
Technology
Technology
 Electronics technology continues to evolve
 18,000 vacuum tubes
 Increased capacity and performance
 Reduced cost
 20 registers, each around 0.5 meters long
 24 meters long
long, 2.5
2 5 meters high
Year
Technology
1951
Vacuum tube
1965
Transistor
1975
I t
Integrated
t d circuit
i it (IC)
1995
Very large scale IC (VLSI)
2005
Ultra large scale IC
Relative performance/cost
1
35
900
2,400,000
6 200 000 000
6,200,000,000
Revolution ?
H
Human civilization
i ili i :
 Agricultural revolution
 Industrial revolution
…
C
Computers
t
are the
th result
lt off an iincredible
dibl
industry that embraces innovations at a
b tht ki rate
breathtaking
t
Yesterday’s
Yesterday s fiction is now reality
 Applications empowered by computers:
Human
Human genome project
 World Wide Web
 Search Engines
Social
Social Networks
 A similar pace in Transportation industry would
h
have
implied
i li d that
th t we reachh N
New Y
Yorkk iin ffew
seconds
Classes of Computers
Classes of Computers
 Desktop computers
 General purpose, variety of software
 Subject to cost/performance tradeoff
 Server computers
 Network based
 High capacity, performance, reliability
 Range from small servers to building sized
 Embedded computers
 Hidden as components of systems
 Stringent power/performance/cost constraints
Below your program
Language of the hardware
 We constantly interact with these computers
 E.g., via apps on iPhone or via a Word
Processor
 The hardware understands on or off.
off Hence,
the language of the hardware needs two
symbols 0 and 1.
1
 How is an application interact with the
hardware?
 Commonly
C
l referred
f
d to as bi
binary digits
di i or bits.
bi
 2 letters do not limit what computers can do
(just like the finite letters in our languages)
(j
g g )
Language of the hardware
 Instructions are collections of bits that the
computer understands
 Hundreds/thousands/millions of lines of code
are hidden
hidd beneath
b
h our apps and
d programs
Hierarchical Layers
y
of Program
g
Code
 High-level language
 Level
L l off abstraction
b
i closer
l
to
problem domain
 Provides for productivity and
portability
 Assemblyy language
g g
 Textual representation of
instructions
 What our programs instruct the computer to
do are the instructions:
 Hardware representation
 Binary digits (bits)
 Encoded instructions and data
Below your program
• Application software
– Written in high-level language
• System software
– Compiler: translates HLL code to
machine code
– Operating System: service code
Under the Covers
 Understanding the underlying hardware (the
computer!) is the main focus of this course
 So, what are the main components of the
computer??
• Handling input/output
• Managing memory and storage
• Scheduling tasks & sharing resources
• Hardware
H d
– Processor, memory, I/O controllers
Components of a Computer
Anatomy of a Computer
 Same components for
allll kinds
ki d off computer
 Desktop, server,
embedded
b dd d
Output device
Network cable
Input
Input device
Input
Input device
Opening the Box
 Motherboard
 I/O connections
Opening the Box
 Memory (DRAM)
 Random access
memory
 Memoryy ((DRAM))
 CPU or central
processing unit
A Safe Place for Data
• Volatile main memory
– Loses
L
iinstructions
t ti
and
dd
data
t when
h power offff
• Non-volatile secondary memory
– Magnetic disk
– Flash memory
– Optical
p
disk ((CDROM, DVD))
AMD Barcelona: 4 processor cores
Abstractions
Inside the Processor (CPU)
 Datapath: performs operations on data
 Control: sequences datapath, memory, ...
 Cache memory
 Small fast SRAM memory for immediate access to
data

Abstraction helps us deal with complexity



Instruction set architecture (ISA)







Instructions
Arithmetic
Th P
The
Processor
Memory
Input/Output
The hardware/software interface
Implementation

Our Topics in this Course
Note abstraction in both hardware and
software
ft
Hide lower-level detail
The details underlying and interface
Performance
 Why is performance important?
 For purchasers: to choose between computers
 For
F d
designers: to make
k the
h sales
l pitchh
 Defining performance is not straightforward!
 An
A analogy
l
with
i h airplanes
i l
shows
h
the
h diffi
difficulty
l
Defining Performance
 Which airplane has the best performance?
Boeing 777
Boeing 777
Boeing 747
Boeing 747
BAC/Sud
Concorde
BAC/Sud
Concorde
Douglas
DC-8-50
Douglas DC8-50
0
100
200
300
400
l
laptop,
the
h one that
h responds
d faster
f
is the
h better!
b
!
Boeing 777
Boeing 777
Boeing 747
Boeing 747
BAC/Sud
Concorde
BAC/Sud
Concorde
Douglas
DC-8-50
Douglas DC8-50
500
1000
2000
4000
6000
 Response time = How long it takes to do a task?
8000 10000
Cruising Range (miles)
Passenger Capacity
0
 Response time
 As a user of a smart phone (embedded computer), or
0
500
Response Time and Throughput
1500
 Response time = the total time required for the computer
0
Cruising Speed (mph)
to complete a task, including disk accesses, memory
accesses, I/O activities,
accesses
activities operating system overheads
overheads, CPU
execution time …
100000 200000 300000 400000
Passengers x mph
Response Time and Throughput
 Throughput
 If I am running a data center with several servers, faster
computer is the one that completes several tasks in one
day!
 Total work done per unit time
• e.g.,
e g tasks/transactions/…
tasks/transactions/ per hour
Relative Performance




Define Performance = 1/Execution Time
Performancex > Performancey
1/Execution Time x > 1/Execution Time y
Execution Time y > Execution Time x
 How are response time and throughput affected by
 Replacing
R l i the
h processor with
i h a faster
f
version?
i ?
 Adding more processors?
 Look
L k in
i the
h textbook
b k for
f a discussion
di
i (P
(Page 28)
 We’ll focus on response time for now…
Performanc e X Performanc e Y
 Execution time Y Execution time X  n
Relative Performance
 Define Performance = 1/Execution Time
 “X is n time faster than Y”
eY
P f
Performanc
e X Performanc
P f
 Execution time Y Execution time X  n

Example: time taken to run a program



10s on A, 15s on B
Execution TimeB / Execution TimeA
= 15s / 10s = 1.5
So A is 1.5 times faster than B
CPU Time
CPU Time  CPU Clock Cycles  Clock Cycle Time

CPU Clock Cycles
C
Clock
Rate
Performance improved
p
byy
 Reducing number of clock cycles
 Increasing clock rate
Measuring Execution Time
 Elapsed
El
d time
i
 Total response time, including all aspects
• Processing,
P
i I/O,
I/O OS overhead,
h d idle
idl time
i
 Determines system performance
 CPU time
ti
 Time spent processing a given job
• Discounts
Di
I/O time,
i
other
h jobs’
j b ’ shares
h
 Comprises user CPU time and system CPU time
 Different programs
r rams are affected differentl
differently by
b CPU
and system performance
CPU Time Example
 Computer A: 2GHz clock, 10s CPU time
 Designing
D i i C
Computer B
 Aim for 6s CPU time
 Can do faster clock,
clock but causes 1.2
1 2 × clock cycles
 How fast must Computer B clock be?
CPU Time B 
CPU Clock CyclesB
Clock RateB
 Hardware designer must often trade off clock rate
against cycle
l count
Clock RateB 
Clock CyclesB 1.2  Clock Cycles A

CPU Time B
6s
CPU Time Example
 Computer A: 2GHz clock, 10s CPU time
 Designing Computer B
 We aim for 6s CPU time
 We
W can do
d faster
f
clock,
l k but
b causes 1.2
1 2 × clock
l k cycles
l
 How fast must Computer B clock be?
CPU Time A 
CPU Clock CyclesA
Clock RateA
Clock CyclesA  CPU Time A  Clock Rate A
 10s  2GHz  20  10 9
Instruction Count and CPI
Clock Cycles  Instruction Count  Cycles per Instruction
CPU Time  Instruction Count  CPI  Clock Cycle Time

Instruction Count  CPI
Clock Rate
CPU Time Example
 Computer A: 2GHz clock, 10s CPU time
 Designing
D i i C
Computer B
 Aim for 6s CPU time
 Can do faster clock,
clock but causes 1.2
1 2 × clock cycles
 How fast must Computer B clock be?
CPU Time B 
Clock RateB 
 Average cycles per instruction
 Determined by CPU hardware
 If different instructions have different CPI
 Average CPI affected by instruction mix
Clock CyclesB 1.2  Clock Cycles A

CPU Time B
6s
1.2  20  10 9 24  10 9
Clock RateB 

 4GHz
6s
6s
CPI Example
 Computer A: Cycle Time = 250ps, CPI = 2.0
 Computer
C
B:
B Cycle
C l Time
Ti
= 500ps,
500 CPI = 11.2
2
 Same ISA
 Which is faster, and by how much?
CPU Time
 Instruction Count for a program
 Determined by program, ISA and compiler
CPU Clock CyclesB
Clock RateB
A
CPU Time
B
 Instruction Count  CPI  Cycle Time
A
A
A is faster…
 I  2.0  250ps  I  500ps
 Instruction Count  CPI  Cycle Time
B
B
 I  1.2  500ps  I  600ps
B  I  600ps  1.2
CPU Time
I  500ps
A
CPU Time
…by this much
Concluding Remarks
Personnel
Cost/performance is improving
Unmesh BordoloiBordoloi Kursledare (examinator)
 Due to underlying technology development
 Kontorstider:
• Vecka 37-43 - Tisdag 14-16
• Epost: unmesh.bordoloi@liu.se
Hierarchical layers
y
of abstraction
 In both hardware and software
Instruction set architecture
 The hardware/software interface
Madeleine Häger Dahlqvist Kurssekreterare
Execution
E
i time:
i
the
h bbest performance
f
measure
Power is a limitingg factor
 Email: madeline.hager.dahlqvist@liu.se
 Use parallelism to improve performance
42
42
Examination
The course will have three obligatory exams/homework
Homework 1
Homework 2
Final Exam
Homework 1
This is obligatory and you can receive 0-4 points. The points
will be added to the final grade.
Homework 2
This is obligatory and you can receive 0-8 points.
points The points
will be added to the final grade.
Final Exam
There
Th
willll be
b a ffinall written grade.
d The
Th points willll be
b added
dd d
to the final grade
43
43
44
44
Examination
Textbook
Final score is calculated as the sum of
 The recommended textbook is:
Homework 1(maximum
(
4 ppoints))
 David A. Patterson and JJohn L. Hennessy:
y
Homework 2 (maximum 8 points)
Computer Organization and Design - The
Hardware / Software Interface,, 4th Edition,,
Morgan Kaufmann
 Available as e-book in the university library
Final Exam (maximalt 28 points)
45
45
Textbook
Other literature
 William Stallings: Computer Organization and
 At the end of every lecture, there will be a
p
covered
slide listingg the relevant portions
from the textbook
Architecture, Prentice Hall International, Inc..
 Sven Eklund, Avancerad datorarkitektur,
Studentlitteratur, 1994.
 Andrew S. Tanenbaum, Structured Computer
Organization, 4th edition, Prentice Hall
International, Inc., 1999.
 V.
V C.
C Hamacher,
Hamacher et al.:
al : Computer Organization,
Organization 4th
edition, McGraw-Hill, 1996.
 Today, we covered 1.1, 1.2, and 1.3
47
46
46
47
48
48