程序代写 CS2305: Computer Architecture

CS2305: Computer Architecture
Fundamentals of Computer Design
(Computer Architecture: Chapter 1 )

Copyright By PowCoder代写 加微信 powcoder

Department of Computer Science and Engineering

Fundamentals Trends in Cost Agenda
 Introduction
 Classes of Computers
 Defining Computer Architecture
 Trends in Technology
 Trends in Power and Energy in ICs  1.6 Trends in Cost
 Dependability
 Measuring Performance
 Quantitative Principles

Fundamentals Trends in Technology
Impact of Time, Volume, Commoditization
 Time: The cost of a computer component decreases over time, why?
 The learning curve!
 Yield: the percentage of manufactured devices that survives the testing procedure
 Volume is another factor in determining cost, why?
 Amortized cost per computer
Microprocessors: price depends on volume, 10% less for each doubling of volume
learning curve

Fundamentals Trends in Technology Impact of Time, Volume, Commoditization
Commodities are products that are sold by multiple vendors in large volumes and are identical
Commoditization means that the market is highly competitive
1) The gap between cost and price is narrowed 2) It helps clearly define a product, and it
increases the competition among the suppliers 4

Fundamentals
Costs of Integrated Circuit (ICs)
 Each copy of the integrated circuit appears in a die
 Multiple dies are placed on each wafer
 After fabrication, the individual dies are separated, tested, and packaged

Fundamentals
Wafer, Die, IC
Processing
F F F F FF F F FF FF F F FF FF F F FF FF F F FF
IC F Packaging F

Fundamentals
Example of Pentium 4
Pentium 4 Processor

Fundamentals
Typical Size of Industrial Wafers

Fundamentals
Integrated Circuit Costs
IC Cost = Die Cost + Testing cost + Packaging Cost Final Test Yield
good dies processed dies
Processing

Fundamentals
Integrated Circuits Costs
IC cost = Die cost + Testing cost + Packaging cost Final test yield
Die cost =
Wafer cost
Dies per Wafer x Die yield
Dies per Wafer =
 ( Wafer_diameter/2)2  ( Wafer_diameter ) Die Area 2 * Die Area 21

Fundamentals
Find the number of dies per 20-cm wafer for a die that is 1.0 cm on a side and a die that is 1.5 cm on a side
Dies per Wafer =
 ( Wafer_diameter/2)2 Die Area
 ( Wafer_diameter ) 2 * Die Area 21

Fundamentals
Integrated Circuit Cost
Where  is a parameter inversely proportional to the number of mask Levels, which is a measure of the manufacturing complexity.
For today’s CMOS process, good estimate is = 11.5-15.5

Fundamentals
Other Costs
Die Test Cost = Test equipment Cost * Ave. Test Time
Packaging Cost: depends on pins, heat dissipation, beauty, …
Die Package
Test & Total assembly
$12 168 PGA $11
Power PC 601
$53 304 QFP $3
HP PA 7100
$73 504 PGA $35
$149 431 PGA $30
Super SPARC
$272 293 PGA $20
$417 273 PGA $19
pins type cost
QFP: Quad Flat Package PGA: Pin Grid Array BGA: Ball Grid Array

Fundamentals Dependability Agenda
 Introduction
 Classes of Computers
 Defining Computer Architecture
 Trends in Technology
 Trends in Power and Energy in ICs  Trends in Cost
 1.7 Dependability
 Measuring Performance
 Quantitative Principles

Fundamentals
Dependability
Dependability
Interruption
Interruption
Service accomplishment
Service accomplishment
Service accomplishmen t
2 Measures:
Mean time to failure (MTTF)
Mean time to repair (MTTR)
Mean time between failures (MTBF) = MTTF + MTTR
Availability = MTTF / MTBF
Module reliability
Module availability

Fundamentals
MTTF of a System
MTTF MTTF MTTF 123
The MTTF of each module is exponentially distributed (independent of age)
Modules are independent of each other What is the MTTF of the system?

Fundamentals
Exponential Distribution and
Poisson Process
Event occurrence
e1e2 e3 t2 t3
 The time, T, between two events is a R.V. following
the exponential distribution
 The arrivals of events follow the Poisson Process. Example, incoming calls of a hotline service

Fundamentals
Characteristics of Poisson
Processe1 Process 1
T2~exp(λ2)
T1~exp(λ1)

Fundamentals
Characteristics of Poisson Process
What process?T3 follows what distribution?
T3~exp(λ3) λ3 = λ1 + λ2

Fundamentals
MTTF of a System
MTTF MTTF MTTF 123
What is the MTTF of the system?
MTTFsystem= 𝟏 𝟏􏰁𝟏􏰁𝟏
MTTF1 MTTF2 MTTF3

Fundamentals
Assumptions: Exponential distributed Failures are independent
MTTF of the whole System?

Fundamentals
How to Increase Dependability?
Redundancy!
 In time ( repeat the operation to see if it is still erroneous)
 In resources (duplicate resources)
200,000-hour MTTF 10-hour MTTR
MTTF of two power supplies?

Fundamentals
Computing the MTTF of Two Suppliers
T1~exp(λ1)
T2~exp(λ2)
MTTF2powers
MTTF2powers= 𝟐 𝑴𝑻𝑻𝑹

Fundamentals Measuring Performance Agenda
 Introduction
 Classes of Computers
 Defining Computer Architecture
 Trends in Technology
 Trends in Power and Energy in ICs  Trends in Cost
 Dependability
 1.8 Measuring Performance
 Quantitative Principles

Fundamentals Measuring Performance Typical Performance Metrics
Response time (or execution time)
• The time between the start and the completion of an event
Throughput
• The total amount of work done in a unit time

Fundamentals
Benchmarks
Benchmark: a common program for testing the execution times of computers
Programs lead to poor performance indication
 Kernels (e.g. matrix multiply)
 Toy programs (e.g. sorting)
 Synthetic benchmarks (e.g. Dhrystone)
Benchmark suit: collection of benchmark programs 26

Fundamentals
SPEC: Standard Performance Evaluation Corporation
 A non-profit corporation formed to establish, maintain and endorse a standardized set of relevant benchmarks that can be applied to the newest generation of high-performance computers.
 SPEC develops benchmark suites and also reviews and publishes submitted results from member organizations and other benchmark licensees.

Fundamentals
Example SPEC Result Report

Fundamentals
Example SPEC Result Report

Fundamentals
Reporting Performance Results
How do we summarize performance, given the execution times of a set of benchmarks?
 Example as shown on the right.

Fundamentals
Option 1: Arithmetic Mean
The arithmetic mean of x1, x2, …, xn is 1n
Meanarith nxi i1
The problem of using arithmetic mean? Benchmark programs with longer execution
times would become more important Example: 4 numbers (5, 6, 5, 7, 100)

Fundamentals
Option 2: Weighted Arithmetic Mean
To add a weighting factor to each benchmark program
The question: how do we set weight factors?
Possible solution: use weights to make programs execute an equal time on a reference computer.
Problem: the reference computer would become crucial

Fundamentals
SPEC Approach: Performance Ratio
Instead of using absolute execution times, use the ratio of performance to a reference computer
A good property of using performance ratio: the selection of reference computer is irrelevant

Fundamentals
SPEC Approach: Geometric Mean (Cont)
The geometric mean of x1, x2, …, xn is n
MeanGeometric  n
Two good properties of geometric mean
1) The geometric mean of the ratios = the ratio of geometric means
2) The choice of the reference computer is irrelevant

Fundamentals Quantitative Principles Agenda
 Introduction
 Classes of Computers
 Defining Computer Architecture
 Trends in Technology
 Trends in Power and Energy in ICs  Trends in Cost
 Dependability
 Measuring Performance
 1.9 Quantitative Principles

Fundamentals
Principles for Computer Design
Take advantage of parallelism
• Data level parallelism and task level parallelism • Pipelining, set-associative caches
• Multicore, multiprocessor, vector
Principle of locality
• Program intends to reuse data and instructions they have used recently
• Temporal locality and spatial locality
Focus on common case
• Favor the frequent case over the infrequent case

Fundamentals
Amdahl’s Law
The law defines the speedup by using a particular feature
Amdahl’s law states that the performance improvement of using a new feature is limited by the fraction of the time the new feature can be used.

Fundamentals Quantitative Principles Processor Performance
DEFINITION:

Fundamentals
Three Factors for Processor Improvement
Clock cycle time
Hardware technology and organization
Organization and instruction set architecture
Instruction count
Instruction set architecture and compiler technology

Fundamentals Quantitative Principles
Different instruction types having different CPIs

程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com