Professional Documents
Culture Documents
Solution
Use of Cache memory
Use of Pipelining Concepts: Overlapping the
execution of instructions
Speed
Data
increases
Storage
Memory
Cach
CPU e Size
increase
Data Hazards
Branch Hazards
Resource Hazards
Flynn’s Taxonomy
SISD single instruction single data stream
SIMD single instruction multiple data stream
MISD multiple instruction single data stream
MIMD Multiple instruction multiple data stream
RISC
Scalability
Large memory
Efficient Throughput
20-30
Slices
processing
steps
Die
Package Dice
Test
Package
Ship
Test
Silicon ingots:
6-12 inches in diameter and about 12-24 inches long
Impurities in the wafer can lead to defective devices and reduces the yield
25
Average cutting edge for non-square pieces
rd s
2
s
Irregular 2
cutting
edge s
2
s
2
pieces rd
2
2 2
s2 s2
rd 2
4 4
2 s2
Total rd 2
No. of 4
Wafer _ diameter
non-square 2 s2
average _ cutting _ length rd 2
pieces 2 2
(Wafer _ Diameter )
s s
rd rd 2 2
2 2
(Wafer _ Diameter )
2 Die _ Area rd 2 2 ( Die _ Area )
rd 2 ( Die _ Area )
26
Integrated Circuits Costs
27
Trends in Cost
Factors that influence the Cost of
Computer
1. Time
2. Volume
3. Commodification
29
Intel Motherboard Components
30
Computer
Components
31
Trends in Cost
Trends in Cost
Cost driven down by learning curve
Yield
Bose-Einstein formula:
agreement
•Module Reliability:- measure of continuous service accomplishment from a
•MTBF:- measure of reliability for repairable system but commonly used for
both repair and non-repair system
If used for repair system
MTBF MTTR MTTF
FIT:- number of expected failures per one billion hours of operation for a
•
device.
34
MTTF Dependability
Operating
properly
Repair
ref t0 t1 t2 t3 t4 t5 t6
downtime
ref
MTTF
uptime/operation uptime
MTTR
downtime
ref
MTBF
36
Dependability
Module reliability
Mean time to failure (MTTF)
Mean time to repair (MTTR)
Mean time between failures (MTBF) = MTTF + MTTR
Module Availability
MTTF
Availabili ty
MTBF
MTTF
Availabilt y
(MTTF MTTR)
Module availability:- measure of the service
accomplishment with respect to the alternation between
the two states of accomplishment and interruption.
37
Real World Examples
38
Performance Metrics
Response (execution) time:
The time between the start and the completion of a task
Measures user perception of the system speed
Common in reactive and time critical systems, single-user computer, etc.
Throughput:
The total number of tasks done in a given time
Most relevant to batch processing (billing, credit card processing)
Mainly used for input/output systems (disk access, printer, etc.)
39
Introduction
Computer Technology
Performance improvements:
Improvements in semiconductor technology
Feature size, clock speed
Improvements in computer architectures
Enabled by HLL compilers, UNIX
Lead to RISC architectures
Trends in Technology
Integrated circuit logic technology
Transistor density: 35%/year
Die size: 10-20%/year
Integration overall: 40-55%/year
Feature size
Minimum size of transistor or wire in x or y
dimension
10 microns in 1971 to .032 microns in 2011
Transistor performance scales linearly
Wire delay does not improve with feature size!
Integration density scales quadratic ally
Dynamic power
½ x Capacitive load x Voltage2 x Frequency switched
Speedup of X relative to Y
Execution timeY / Execution timeX
Execution time
Wall clock time: includes all system overheads
CPU time: only computation time
Benchmarks
Kernels (e.g. matrix multiply)
Toy programs (e.g. sorting)
Synthetic benchmarks (e.g. Dhrystone)
Benchmark suites (e.g. SPEC06fp, TPC-C)
Principle of Locality
Reuse of data and instructions
57
Question: Suppose that we want to enhance the
processor used for web serving. The new
processor is 10 times faster on computation in
the web serving application than the original
processor. Assuming that the original
processor is busy with computation 40% of the
time and is waiting I/O 60% of the time, what is
the overall speedup gained by incorporating
the enhancement?
59
Question??
Q2. Find the die yield for dies that are 1.5 cm on a
side and 1.0 cm on a side, assuming a defect
density of 0.4 per cm2 and α is 4.
60
f1 = 500 MHz f2 = 2.5 GHz
T1 = 12x seconds T2 = x seconds
MIPS Rate1 = 100 MIPS MIPS Rate2 = 1800 MIPS
Copyright © 2012, Elsevier Inc. All rights reserved. 61
CPI1 = ? CPI2 = ?
Ic = ? Ic = ?
Throughput1 = ? Throughput2 = ?
Program 1 1 10 20
Using the simplifying assumptions that the lifetimes are exponentially distributed and that
failures are independent, compute the MTTF of the system as a whole.
1 1 1 1 1
10
Failure rate system 1000,000 500000 200000 200000 1000000
10 2 5 5 1 23 23 1000
1,000,000 hours 1,000,000 1000,000,000 hours
23,000
1 billion hours
or, 23000 FIT
1 1,000,000,000 hours
MTTFsystem 43,500 hours
Failure rate system 23,000
1 years 364 24 hours 8736 hours
43500
Therefore, MTTFsystem 4.979 years
8736
64
Q2. Availability is the most important consideration
for designing servers, followed closely by scalability
and throughput.
(a)We have a single processor with a failures in time (FIT) of
100. What is the mean time to failure (MTTF) for this system?
(b)If it takes 1 day to get the system running again, what is the
availability of the system?
(c)Imagine that the government, to cut costs, is going to build a
supercomputer out of inexpensive computers rather than
expensive, reliable computers. What is the MTTF for a system
with 1000 processors? Assume that if one fails, they all fail.
65
1
a) MTTFsystem
Failure ratesystem
1,000,000,000 hours
100
107 hours
67
1
c ) Failure ratesystem 1000 7
10
1000 100
107 100
100,000
9
10 hours
100000
1 billion hours
100,000 FIT
1000,000,000
MTTFsystem 10,000 hours
100,000
68