CPU Scheduling

Basic Concepts
 Maximum CPU utilization
obtained with
multiprogramming

 CPU–I/O Burst Cycle –

Process execution consists
of a cycle of CPU
execution and I/O wait

 CPU burst followed by

I/O burst

 CPU burst distribution is of

main concern
Histogram of CPU-burst Times
CPU Scheduler
 Short-term scheduler selects from among the
processes in ready queue, and allocates the CPU to one of
them
 Queue may be ordered in various ways
 CPU scheduling decisions may take place when a process:
1. Switches from running to waiting state
2. Switches from running to ready state
3. Switches from waiting to ready
4. Terminates
 Scheduling under 1 and 4 is non-preemptive
 All other scheduling is preemptive
 Consider access to shared data
 Consider preemption while in kernel mode
 Consider interrupts occurring during crucial OS activities
Preemptive Vs. Non Preemptive Scheduling
Dispatcher
 Dispatcher module gives control of the CPU to
the process selected by the short-term
scheduler; this involves:
◦ switching context
◦ switching to user mode
◦ jumping to the proper location in the user program
to restart that program

 Dispatch latency – time it takes for the

dispatcher to stop one process and start
another running
Scheduling Criteria
 CPU utilization – keep the CPU as busy as possible

 Throughput – # of processes that complete their execution

per time unit

 Turnaround time – amount of time to execute a particular

process
 Turnaround time (TAT)=Completion time – Arrival time

 Waiting time – amount of time a process has been waiting

in the ready queue

 Response time – amount of time it takes from when a

request was submitted until the first response is produced,
not output (for time-sharing environment)
Scheduling Algorithm Optimization
Criteria

 Max CPU utilization

 Max throughput
 Min turnaround time
 Min waiting time
 Min response time
First-Come, First-Served (FCFS) Scheduling

Process Burst Time

P1 24
P2 3
P3 3
 Suppose that the processes arrive in the
order: P1 , P2 , P3

The Gantt PChart for the schedule
1 P
2 P is:
3

0 24 27 30

 Waiting time for P1 = 0; P2 = 24; P3 = 27

 Average waiting time: (0 + 24 + 27)/3 = 17
 FCFS Scheduling (Cont.)

Suppose that the processes arrive in the order:

P2 , P3 , P1
 The Gantt chart for the schedule is:

P2 P3 P1

0 3 6 30

 Waiting time for P1 = 6; P2 = 0; P3 = 3

 Average waiting time: (6 + 0 + 3)/3 = 3
 Much better than previous case
 Convoy effect - short process behind long process
◦ Consider one CPU-bound and many I/O-bound processes
Shortest-Job-First (SJF) Scheduling

 Associate with each process the length of

its next CPU burst
◦ Use these lengths to schedule the process with
the shortest time

 SJF is optimal – gives minimum average

waiting time for a given set of processes
◦ The difficulty is knowing the length of the next
CPU request
◦ Could ask the user
Example of SJF
ProcessArriva l Time Burst Time
P1 0.0 6
P2 2.0 8
P3 4.0 7
P4 5.0 3
 SJF scheduling chart
P4 P1 P3 P2

0 3 9 16 24

 Average waiting time = (3 + 16 + 9 + 0) / 4 = 7

Shortest Remaining Time First
Scheduling

Preemptive version of SJN

Whenever a new process arrives in the ready queue, the decision

on which process to schedule next is redone using the SJN
algorithm.

Is SRT more “optimal” than SJN in terms of the minimum average

waiting time for a given set of processes?
Example of Shortest-remaining-
time-first
 Now we add the concepts of varying arrival times and preemption to
the analysis
Process Arrival Time Burst Time
P1 0 8
P2 1 4
P3 2 9
P4 3 5
 Preemptive SJF Gantt Chart

P1 P2 P4 P1 P3
0 1 5 10 17 26

 Average waiting time = [(10-1)+(1-1)+(17-2)+(5-3)]/4 =

26/4 = 6.5
Round Robin (RR)
 Each process gets a small unit of CPU time (time quantum q),
usually 10-100 milliseconds. After this time has elapsed, the
process is preempted and added to the end of the ready queue.
 If there are n processes in the ready queue and the time
quantum is q, then each process gets 1/n of the CPU time in
chunks of at most q time units at once. No process waits more
than (n-1)q time units.
 Timer interrupts every quantum to schedule next process
 Performance
◦ q large  FIFO (FCFS)
◦ q small  RR
 Note that q must be large with respect to context switch,
otherwise overhead is too high
 Process Time
P1 24
P2 3
P3 3
 The Gantt chart is:

 Typically, higher average turnaround than SJF, but better

response
 q should be large compared to context switch time
◦ q usually 10 milliseconds to 100 milliseconds,
◦ Context switch < 10 microseconds
Determining Length of Next CPU Burst

 Can only estimate the length – should be similar

to the previous one
◦ Then pick process with shortest predicted next CPU
burst
 Can be done by using the length of previous
CPU bursts, using exponential averaging

 Commonly, α set to ½
Shortest Remaining Time First
Scheduling

Preemptive version of SJN

Whenever a new process arrives in the ready

queue, the decision on which process to
schedule next is redone using the SJN
algorithm.

Is SRT more “optimal” than SJN in terms of the

minimum average waiting time for a given set
of processes?
 Example of Shortest-remaining-time-first

 Now we add the concepts of varying arrival times and

preemption to the analysis
Process i Arrival TimeT Burst Time
P1 0 8
P2 1 4
P3 2 9
P4 3 5
 Preemptive SJF Gantt Chart

P1 P2 P4 P1 P3
0 1 5 10 17 26

 Average waiting time = [(10-1)+(1-1)+(17-2)+(5-3)]/4

= 26/4 = 6.5
 Round Robin (RR)
 Each process gets a small unit of CPU time (time
quantum q), usually 10-100 milliseconds. After this
time has elapsed, the process is preempted and added
to the end of the ready queue.
 If there are n processes in the ready queue and the time
quantum is q, then each process gets 1/n of the CPU
time in chunks of at most q time units at once. No
process waits more than (n-1)q time units.
 Timer interrupts every quantum to schedule next process
 Performance
◦ q large  FIFO (FCFS)
◦ q small  RR
 Note that q must be large with respect to context switch,
otherwise overhead is too high
Example of RR with Time Quantum = 4

Process Burst Time

P1 24
P2 3
P3 3
 The Gantt chart is:
P1 P2 P3 P1 P1 P1 P1 P1
0 4 7 10 14 18 22 26 30

 Typically, higher average turnaround than SJF, but

better response

 q should be large compared to context switch time

◦ q usually 10 milliseconds to 100 milliseconds,
◦ Context switch < 10 microseconds
Time Quantum and Context Switch Time
Turnaround Time Varies With The Time
Quantum

80% of CPU bursts

should be shorter
than q
 Priority Scheduling

A priority number (integer) is associated with each process

The CPU is allocated to the process with the highest priority (smallest integer
 highest priority)

• Preemptive
• Nonpreemptive

SJF is priority scheduling where priority is the inverse of predicted next CPU
burst time

Problem  Starvation – low priority processes may never execute

Solution  Aging – as time progresses increase the priority of the process

Example of Priority Scheduling

Process Burst Time Priority

P1 10 3
P2 1 1
P3 2 4
P4 1 5
P5 5 2

 Priority scheduling Gantt Chart

 Average waiting time = 8.2

Priority Scheduling w/ Round-Robin
 Run the process with the highest priority.
Processes with the same priority run round-
robin
 Example:
Process a Burst Time Priority
P1 4 3
P2 5 2
P3 8 2
P4 7 1
P5 3 3
 Gantt Chart with time quantum = 2
 Multilevel Queue

 The ready queue consists of multiple

queues
 Multilevel queue scheduler defined

by the following parameters:

◦ Number of queues
◦ Scheduling algorithms for each queue
◦ Method used to determine which queue a
process will enter when that process needs
service
◦ Scheduling among the queues
 Multilevel Queue

 With priority
scheduling, have
separate queues for
each priority.
 Schedule the

process in the
highest-priority
queue!
 Multilevel Queue

 Prioritization based upon process type

 Multilevel Feedback Queue
 A process can move between the various
queues.
 Multilevel-feedback-queue scheduler defined by
the following parameters:
◦ Number of queues
◦ Scheduling algorithms for each queue
◦ Method used to determine when to upgrade a process
◦ Method used to determine when to demote a process
◦ Method used to determine which queue a process will
enter when that process needs service
 Aging can be implemented using multilevel
feedback queue
 Example of Multilevel Feedback Queue
• Q0 – RR with time quantum 8
milliseconds
Three • Q1 – RR time quantum 16
queues: milliseconds
• Q2 – FCFS

• A new process enters queue Q0 which

is served in RR
• When it gains CPU, the process
receives 8 milliseconds
• If it does not finish in 8 milliseconds,
Schedulin
the process is moved to queue Q1
g • At Q1 job is again served in RR and
receives 16 additional milliseconds
• If it still does not complete, it is
preempted and moved to queue Q2
 Priority-based Scheduling

 For real-time scheduling, scheduler must support

preemptive, priority-based scheduling
◦ But only guarantees soft real-time
 For hard real-time must also provide ability to
meet deadlines
 Processes have new characteristics: periodic
ones require CPU at constant intervals
◦ Has processing time t, deadline d, period p
◦0≤t≤d≤p
◦ Rate of periodic task is 1/p
Operating System Examples

Linux scheduling

Windows scheduling

Solaris scheduling
 Linux Scheduling Through Version 2.5

 Prior to kernel version 2.5, ran variation of standard UNIX scheduling algorithm
 Version 2.5 moved to constant order O(1) scheduling time
◦ Preemptive, priority based
◦ Two priority ranges: time-sharing and real-time
◦ Real-time range from 0 to 99 and nice value from 100 to 140
◦ Map into global priority with numerically lower values indicating higher priority
◦ Higher priority gets larger q
◦ Task run-able as long as time left in time slice ( active)
◦ If no time left (expired), not run-able until all other tasks use their slices
◦ All run-able tasks tracked in per-CPU runqueue data structure
 Two priority arrays (active, expired)
 Tasks indexed by priority
 When no more active, arrays are exchanged
◦ Worked well, but poor response times for interactive processes
 Linux Scheduling in Version 2.6.23 +

 Completely Fair Scheduler (CFS)

 Scheduling classes
◦ Each has specific priority
◦ Scheduler picks highest priority task in highest
scheduling class
◦ Rather than quantum based on fixed time allotments,
based on proportion of CPU time
◦ Two scheduling classes included, others can be
added
1.default
2.real-time
 Linux Scheduling in Version 2.6.23 +
(Cont.)
 Quantum calculated based on nice value from
-20 to +19
◦ Lower value is higher priority
◦ Calculates target latency – interval of time during
which task should run at least once
◦ Target latency can increase if say number of active
tasks increases
 CFS scheduler maintains per task virtual run
time in variable vruntime
◦ Associated with decay factor based on priority of task
– lower priority is higher decay rate
◦ Normal default priority yields virtual run time =
actual run time
 To decide next task to run, scheduler picks
CFS Performance
 Linux Scheduling (Cont.)

 Real-time scheduling according to POSIX.1b

◦ Real-time tasks have static priorities
 Real-time plus normal map into global priority
scheme
 Nice value of -20 maps to global priority 100
 Nice value of +19 maps to priority 139
 Linux Scheduling (Cont.)
 Linux supports load balancing, but is also
NUMA-aware.
 Scheduling domain is a set of CPU cores that

can be balanced against one another.

 Domains are organized by what they share (i.e.,

cache memory.) Goal is to keep threads from

migrating between domains.
 Windows Scheduling
 Windows uses priority-based preemptive
scheduling
 Highest-priority thread runs next
 Dispatcher is scheduler
 Thread runs until (1) blocks, (2) uses time slice,
(3) preempted by higher-priority thread
 Real-time threads can preempt non-real-time
 32-level priority scheme
 Variable class is 1-15, real-time class is 16-31
 Priority 0 is memory-management thread
 Queue for each priority
 If no run-able thread, runs idle thread
Windows Priorities
 Solaris

 Priority-based scheduling
 Six classes available
◦ Time sharing (default) (TS)
◦ Interactive (IA)
◦ Real time (RT)
◦ System (SYS)
◦ Fair Share (FSS)
◦ Fixed priority (FP)
 Given thread can be in one class at a time
 Each class has its own scheduling algorithm
 Time sharing is multi-level feedback queue
◦ Loadable table configurable by sysadmin