Scribd
Upload
131 views13 pages

15 Performance Metrics

Aperiodic, periodic and sporadic tasks need to be taken into account separately. An RTOS should quickly and predictably respond to the event. Interrupt and task execution latencies with respect to the sum of the execution times must be very small.

Uploaded by

dugguz
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
131 views13 pages

15 Performance Metrics

Aperiodic, periodic and sporadic tasks need to be taken into account separately. An RTOS should quickly and predictably respond to the event. Interrupt and task execution latencies with respect to the sum of the execution times must be very small.

Uploaded by

dugguz
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

REAL TIME OPERATING SYSTEMS

Lesson-23:
Performance Metrics

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 1
Inc.
1. Models for Performance Measures

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 2
Inc.
Three types of tasks for finding
performance
 Scheduler must take into account (aperiodic,
periodic and sporadic) separately.
 (i) An aperiodic task needs to be run only once.
 (ii) A periodic task needs to run after the fixed
periods and it that must be executed before its next
preemption is needed.
 (iii) A sporadic task needs to be checked for run
after a minimum time period of its occurrence.

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 3
Inc.
Predictably response to the event and
minimum interrupt latency as Performance
Measures
 An RTOS should quickly and
predictably respond to the event.
 It should minimum interrupt latency
and fast context switching latency.

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 4
Inc.
Three Models for Performance Measures

 (i) Ratio of the sum of latencies of the


tasks and Interrupt with respect to the
sum of the execution times.
 (ii) CPU load for how much time
CPU not idle
 (iii) Worst-Case Execution time with
respect to mean execution time.

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 5
Inc.
Interrupt latencies as Performance Metric

 Interrupt and task execution latencies


with respect to the sum of the
execution times must be very small
 There must be fast context switching .

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 6
Inc.
CPU Load as Performance Metric
 Each task gives a load to the CPU
that equals the task execution time
divided by the task period
 CPU load or system load estimation in the
case of multitasking is as follows.
Suppose there are m tasks. For the
multiple tasks, the sum of the CPU loads
for all the tasks and ISRs should be less
than 1

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 7
Inc.
CPU Load

 CPU load equal to 0.1 (10%)─ means the


CPU is underutilized and spends its 90%
time in a waiting mode.
 Since the executions times can vary or and
the task periods vary, the CPU loads can
also vary

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 8
Inc.
Sporadic Task Model Performance Metric

 Ttotal = Total length of periods for which


sporadic tasks occur
 e = Total Task Execution Time
 Tav = Mean periods between the sporadic
occurrences
 Tmin = Minimum Period between the
sporadic occurrences

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 9
Inc.
Sporadic Task Model Performance Metric

 Worst-Case Execution-time performance


metric, p is measured calculated as follows
for a tasks worst case of a task in a model.
model.
 p = p worst= (e * Ttotal / Tav )/ (e * Ttotal / Tmin).
 Because average rate of occurrence of
sporadic task = (Ttotal / Tav) and maximum
rate of sporadic task burst = Ttotal / Tmin.

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 10
Inc.
Summary

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 11
Inc.
We learnt
 Various models to define a performance
metric.
 Three performance metrics for schedule
management by the RTOS
 (i) interrupt latencies with respect to the
execution times
 (ii) CPU load.
 (iii) Worst case execution time.

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 12
Inc.
End of Lesson 23 of Chapter 8

Chapter-8 L23: "Embedded Systems - Architecture,


2008 Programming and Design" , Raj Kamal, Publs.: McGraw-Hill, 13
Inc.

You might also like