INTERRUPT LATENCY AND RESPONSE OF THE TASK

1. INTERRUPT LATENCY AND
RESPONSE OF THE TASK
An RTOS should quickly and predictable
respond of the event.
 It should have minimum interrupt
latency and fast context switching
latency.
 Different models have been proposed
for measuring performance.
1/25/2015 J.SUDARSHANREDDY

2. There are three components of
the performance of processing as
follows.
1.Ratio of the sum of interrupt latencies with
respect to the sum of the execution times.
2.Cpu load.
3.Worst-case execution times with respect to
the mean execution time.
1/25/2015 J.SUDARSHANREDDY

3.  Interr Ratio of the sum of interrupt
latencies with respect to sum of the
execution time.
 CPU latencies in various task models can be
used for evaluating performance metrics.
 The latencies for various task scheduling
models are described.
 The CPU load is another way to look at the
performance
 Worst case performance calculation for a
sporadic task is performed
1/25/2015 J.SUDARSHANREDDY

4.  In real time systems by Jane W. S. Liu.
Pearson Education, 2000, for details of
many models available for evaluating
the performance.
CPU load as performance
metric
 Each task gives a load to the CPU
that execution time divided by the
task period.
 Recall in_Aout_B intranet work
1/25/2015 J.SUDARSHANREDDY

5.  EX: Receiver port A expects another
character before 172ms.
 task period is 172ms,
 if the task execution time is 172ms.
 the CPU load for this taskis1(100%).
 The task execution time when a
character is received must be than
172ms.
 The maximum load of the cpu is1.
1/25/2015 J.SUDARSHANREDDY

6.  The CPU load having multitasking as
follow
 The sum of the CPU loads for all the
tasks and ISR should be less than 1.
 The time outs and fixed time limit
definitions for the task reduce the CPU
load.
 CPU load for the higher-priority tasks.
 So that even the lower-priority tasks can
be run before the dead lines.
1/25/2015 J.SUDARSHANREDDY

7.  The some of the CPU loads equal to
0.1(10%)!
 the CPU is under utilized and spends
90% of its time in a waiting mode.
 The execution and task time is vary
CPU loads also vary.
 A task needs to run only Once.
 Then it is aperiodic in an application.
 Scheduling of the task that need to run
periodically.
1/25/2015 J.SUDARSHANREDDY

8.  The fixed periods can be periodic and
CPU load very close to 1.
 EX: inputs as a port with
predetermined periods and the inputs
are in succession without any time
gap.
 A task can’t be scheduled at fixed
periods. its schedule is called
sporadic.
 EX: if a task is expected to receive
1/25/2015 J.SUDARSHANREDDY

9. Inputs at variable timegaps
then the task schedule is
sporadic.
 The variable time gaps must be with
defined limits.
 A preemptive scheduler must taken into
account three types of tasks.
 There are:
1.Aperiodic
2.Periodic
3.sporadic
1/25/2015 J.SUDARSHANREDDY

10.  Aperiodic: the task needs to be
preempted only once.
 Periodic: the task needs to be
preempted after the fixed periods and it
must be executed before its next
preemption is needed.
 Sporadic: the task needs to be checked
after a minimum time period of its
occurrence.
 The software designers to keep the CPU
load between(0.7 +or- 0.25)for sporadic
tasks.
1/25/2015 J.SUDARSHANREDDY

11. Worst case execution time
performance
T total is the total length of the
periods for which sporadic task
occur.
E is the total task execution
time.
TAV is the mean periods
between the sporadic
occurrence.
1/25/2015 J.SUDARSHANREDDY

12.  Worst case execution time performance
metric, p is calculated as follows.
 The worst case task in a model is
P=P worst=(e*T total / T av)/(e*T total / Tmin)
The average rate of sporadic task
(Ttotal/ T av).
The maximum rate of sporadic task burst
is Ttotal/Tmin.
1/25/2015 J.SUDARSHANREDDY