0% found this document useful (0 votes)

22 views

ch02.3 - CPU Scheduling

This document discusses CPU scheduling in operating systems. It describes the long term scheduler, short term scheduler, and medium term scheduler. It also covers basic CPU scheduling concepts like single program vs multiprogramming. It analyzes scheduling criteria and different scheduling algorithms like preemptive vs nonpreemptive.

Uploaded by

22110187

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

22 views

ch02.3 - CPU Scheduling

Uploaded by

22110187

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 51

Chapter 2: Process Management

2.3. CPU Scheduling

GV: Nguyễn Thị Thanh Vân

Objectives

 Describe various CPU scheduling algorithms

 Assess CPU scheduling algorithms based on scheduling criteria
 Explain the issues related to multiprocessor and multicore scheduling
 Describe various real-time scheduling algorithms
 Describe the scheduling algorithms used in the Windows, Linux, and
Solaris operating systems
 Apply modeling and simulations to evaluate CPU scheduling
algorithms

1
Outline
 Basic Concepts
 Scheduling Criteria
 Scheduling Algorithms

Scheduling in OS

 Long term scheduler - also known as job scheduler.

• It chooses the processes from the pool (secondary memory) and keeps
them in the ready queue maintained in the primary memory.
• It mainly controls the degree of Multiprogramming.
• to choose a perfect mix of IO bound and CPU bound processes among
the jobs present in the pool.
 Short term scheduler - also known as CPU scheduler.
• It selects one of the Jobs from the ready queue and dispatch to the
CPU for the execution.
• A scheduling algorithm is used to select which job is going to be
dispatched for the execution.
 Medium term scheduler - takes care of the swapped out processes.
• If the running state processes needs some IO time for the completion
then there is a need to change its state from running to waiting.

2
Basic Concepts
 Single program: A process is executed until it must
wait for some I/O request.
• => the CPU then just sits idle.
• waiting time is wasted; no useful work is
accomplished
 Multiprogramming: to maximize CPU utilization.
• Processes are kept in memory at one time.
• When 1 process has to wait, the OS takes the
CPU away from that process and gives the CPU
to another process.
• => keeping the CPU busy is extended to all
processing cores on the system.

 CPU scheduling: central to OS design

• Almost all computer resources are scheduled
• CPU is one of the primary

Basic Concepts - CPU–I/O Burst Cycle

 CPU scheduling: Process execution consists of
• a cycle of CPU execution and I/O wait
 Processes have 2 states (alternate): in fig
• CPU burst followed by I/O burst
 Excution
• begins with a CPU burst.,
• I/O burst
• …
• the final CPU burst ends with a system
request to terminate execution
 The durations of CPU bursts have been
measured extensively

3
Histogram of CPU-burst Times
 CPU burst distribution is of main concern
• An I/O-bound program typically has many short CPU bursts.
• A CPU-bound program might have a few long CPU bursts.

 A large number of short CPU bursts and a small number of long CPU
bursts.
 This distribution can be important when implementing a CPU-scheduling

CPU Scheduler
 The CPU scheduler selects from among the processes in ready queue,
and allocates a CPU core to one of them
• The ready queue may be ordered in various ways: FIFO, FCFS, SJF
• The records in the queues are generally process control blocks (PCBs)
of the processes
 CPU scheduling decisions may take place when a process:
1. Switches from running to waiting state => need schedule
2. Switches from running to ready state => a choice (an option; a decision)
3. Switches from waiting to ready state => a choice
4. Terminates => need schedule

4
Preemptive and Nonpreemptive Scheduling
 For situations 1 and 4, there is no choice in terms of scheduling. A new process
(in the ready queue) must be selected for execution:
• nonpreemptive or cooperative.
 For situations 2 and 3, however, there is a choice . (an option; a decision)
• Preemptive
 Nonpreemptive (không ưu tiên trước - Điều phối độc quyền )
• is used when a process terminates, or switches from running to the waiting.
• once the CPU cycles are allocated to a process, the process holds the CPU
till it gets terminated or reaches a waiting state.
 Preemptive (có ưu tiên - Điều phối không độc quyền)
• is used when a process switches from running / waiting to ready.
• The CPU cycles are allocated to the process for a limited amount of time
and then taken away, and the process is again placed back in the ready
queue to get its next chance to execute.
 Virtually all modern operating systems including Windows, MacOS, Linux, and
UNIX use preemptive scheduling algorithms.
Operating System Concepts – 10th Edition 5a.9 Silberschatz, Galvin and Gagne ©2018

Preemptive and Nonpreemptive Scheduling

Para-
Preemptive scheduling non-preemptive scheduling
meter
Once resources (CPU Cycle) are allocated to a
In this resources (CPU Cycle) are allocated
Basic process, the process holds it till it completes its
to a process for a limited time.
burst time or switches to waiting state.

Process can not be interrupted until it

Interrupt Process can be interrupted in between.
terminates itself or its time is up.
If a process having high priority frequently If a process with a long burst time is running
Starvation arrives in the ready queue, a low priority CPU, then later coming process with less CPU
process may starve. burst time may starve.
It has overheads of scheduling the
Overhead It does not have overheads.
processes.
Flexibility flexible Rigid (inflexible)
Cost cost associated no cost associated
CPU
CPU utilization is high. It is low in non preemptive scheduling.
Utilization
Round Robin and First Come First Serve and
Examples
Shortest Remaining Time First. Shortest Job First.

5
Preemptive Scheduling and Race Conditions

 Preemptive scheduling can result in race conditions when data are shared
among several processes.
 Consider the case of two processes that share data (Race Conditions).
• While one process is updating the data, it is preempted so that the
second process can run.
• The second process then tries to read the data, which are in an
inconsistent state.
 We saw this in the bounded buffer example
 This issue will be explored in detail in Chapter 6.

Dispatcher
 Dispatcher module gives control of the CPU to
the process selected by the CPU 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

6
Scheduling Criteria

Max  CPU utilization – keep the CPU as busy as possible

Max  Throughput – # of processes that complete their execution per time

unit

Min  Turnaround time – amount of time to execute a particular process

Min  Waiting time – amount of time a process has been waiting in the
ready queue

Min  Response time – amount of time it takes from when a request was
submitted until the first response is produced.
Optimization Criteria for Scheduling Algorithms

Calculating the times

7
Outline
 Basic Concepts
 Scheduling Criteria
 Scheduling Algorithms

Scheduling Algorithms

1. First-Come, First-Served Scheduling

2. Shortest-Job-First Scheduling
3. Round-Robin Scheduling
4. Priority Scheduling (version Preemptive + Non- Preemptive)
5. Multilevel Queue Scheduling
6. Multilevel Feedback Queue Scheduling

8
First- Come, First-Served (FCFS) Scheduling
 FCFS: the process that requests the CPU first is allocated the CPU first
 The implementation of the FCFS is easily managed with a FIFO queue.
• When a process enters the ready queue,
• its PCB is linked onto the tail of the queue.
• When the CPU is free, it is allocated to the process at the head of the
queue.
• The running process is then removed from the queue.

 The average waiting time under the FCFS policy is often quite long.

First- Come, First-Served (FCFS) Scheduling

 Example with 3 processes Process Burst Time

P1 24
P2 3
P3 3
 Suppose that the processes arrive in the order: P1 , P2 , P3
The Gantt Chart for the above schedule is:
P1 P2 P3
0 24 27 30

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

 Average waiting time: (0 + 24 + 27)/3 = 17
 Operation: Once the CPU has been allocated to a process => it keeps the
CPU until it releases the CPU, either by terminating or by requesting I/O.
 The FCFS particularly troublesome for interactive systems, where it is
important that each process get a share of the CPU at regular intervals.
• FCFS: Nonpreemptive

9
Scheduling Algorithms

1. First-Come, First-Served Scheduling

2. Shortest-Job-First Scheduling
3. Round-Robin Scheduling
4. Priority Scheduling
5. Multilevel Queue Scheduling
6. Multilevel Feedback Queue Scheduling

FCFS Scheduling (Cont.)

Suppose that the processes arrive in the order: P2 , P3 , P1 . The Gantt chart :
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
 FCFS may suffer from the convoy effect
• if the burst time of the first job is the highest among all then the processes
of lower burst time may get blocked
 they may never get the CPU if the job in the execution has a very high burst time.

10
FCFS Scheduling (Cont.)
 Ex, Consider one CPU-bound and many I/O-bound processes
• The CPU-bound process will get and hold the CPU, all the other processes will
finish their I/O and will move into the ready queue, waiting for the CPU.
=> I/O devices are idle.
• the CPU-bound process finishes its CPU burst and moves to an I/O device. All the
I/O-bound processes, execute quickly and move back to the I/O queues
=> CPU sits idle.
 Hence in Convoy Effect, one slow process slows down the performance of
the entire set of processes, and leads to wastage of CPU time and other
devices.
 To avoid Convoy Effect, preemptive scheduling algorithms like Round Robin
Scheduling can be used
• as the smaller processes don’t have to wait much for CPU time – making
their execution faster and leading to less resources sitting idle.

Shortest-Job-First (SJF) Scheduling

 SJF associates with each process the length of its next CPU burst
• Use these lengths to schedule the process with the shortest time
• If the next CPU bursts of two processes are the same, FCFS
scheduling is used to break the tie
 SJF is optimal – gives minimum average waiting time for a given set of
processes:
• Moving a short price before a long one
 How do we determine the length of the next CPU burst?
• Could ask the user
• Estimate

11
Shortest-Job-First (SJF) Scheduling
 The SJF algorithm can be either preemptive or nonpreemptive.
• Non-preemptive SJF algorithm will allow the currently running process to
finish its CPU burst.
• A preemptive SJF algorithm will preempt the currently executing process,
(a new process arrives with less work than the remaining time of currently executing proc:
=> will run)
 Moving a short process before a long one decreases the waiting time of the short
process more than it increases the waiting time of the long process. Consequently, the
average waiting time decreases.

P2 P0 P3 P1 P2 P3 P0 P1 P2
0 6 8 12 16 0 1 5 7 11 16

Detail, later =>

Shortest Remaining Time First (SRTF)

 Preemptive SJF: Shortest-remaining-time-first scheduling (SRTF):
• Whenever a new process arrives in the ready queue, the decision on which
process to schedule next is redone using the SJN (nonpreemtive SJF)
• Is SRTF more “optimal” than SJN in terms of the minimum average waiting
time for a given set of processes?

• Ưu tiên: arri sớm + CPU time nhỏ

• 0: P2 arr&run
• 1: P3 arr, CPU time: P3(4)<P2(5) => P3 run trc
• 2: P1 arr, CPU time: P3(3)<P1(4), P2(5) => P3
run tiếp
• 3: P0 arr, CPU time: P3(2)=P0(2), P2(5), P1(4)
=> P3 run tiếp vì arr trc (sameP1)
• 5: P3 done. P0(2)< P2(5), P1(4) => P0 run trc
• 7: P0 done. P1(4)<P2(5) -> P1 run trc
• 11: P1 done. P2 run
• 16: P2 done
• AWT?
Operating System Concepts – 10th Edition 5a.24 Silberschatz, Galvin and Gagne ©2018

12
Example of SRTF (Preemption SJF)
 Preemption SJF to the analysis - arri early + CPU time small
ProcessA aArrival TimeTBurst Time
P1 0 8
P2 1 4
P3 2 9
P4 3 5
 Analysis: Trace
time 0 1 2 3 5 10 17 26
P1 8 7 7 7 7 7 0 0
P2 4 3 2 0 0 0 0
P3 9 9 9 9 9 0
P4 5 4 0 0 0
 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 = 6.5ms

Example of NonPreemptive SJF

 Non Preemption to the analysis – no break
ProcessA Arrival TimeTBurst Time
P1 0 8
P2 1 4
P3 2 9
P4 3 5
 Analysis: time 0 1 2 3 5 8 12 17 26
P1 8 7 6 5 3 0 0 0 0
P2 4 4 4 4 4 0 0 0
P3 9 9 9 9 9 9 0
P4 5 5 5 5 0 0
 Gantt Chart: P1 P2 P4 P3
0 8 12 17 26

 Average waiting time of 7.75 milliseconds = (0+ (8-1)+(12-3)+(17-2))/4

 FCFS result in an average waiting time 8.75= (0+7+10+18)/4
P1 P2 P3 P4
Operating System Concepts – 10th Edition 5a.26 Silberschatz, Galvin and Gagne ©2018

13
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

 n 1   t n  1    n .
 Equation:

• n+1 = giá trị dự đoán cho thời gian sử dụng CPU tiếp sau
• tn = thời gian thực tế của sự sử dụng CPU thứ n
• , 0    1
• 0 là một hằng số

Prediction of the Length of the Next CPU Burst

Figure shows an exponential average with 𝛼 = ½ and τ0 = 10

14
Examples of Exponential Averaging

  =0
 n 1   t n  1    n .
• n+1 = n = 0
• Recent history does not count
  =1
• n+1 =  tn = tn
• Only the actual last CPU burst counts
 If we expand the formula, we get:
n+1 =  tn+(1 - ) tn -1 + …
+(1 -  )j  tn -j + …
+(1 -  )n +1 0

 Since both  and (1 - ) are less than or equal to 1, each successive

term has less weight than its predecessor

Scheduling Algorithms

1. First-Come, First-Served Scheduling

2. Shortest-Job-First Scheduling
3. Round-Robin Scheduling
4. Priority Scheduling
5. Multilevel Queue Scheduling
6. Multilevel Feedback Queue Scheduling

15
Round Robin (RR)
 RR: similar to FCFS scheduling, but preemption is added to enable the
system to switch between processes
 Each process gets a small unit of CPU time (time quantum q - định lượng
thời gian),
• 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

Example of RR with Time Quantum = 4

Process Burst Time
P1 24
P2 3
P3 3
 The Gantt chart is: order P1,P2,P3
P1 P2 P3 P1 P1 P1 P1 P1
0 4 7 10 14 18 22 26 30

 Waiting time for

• P1: 3(p2)+3(p3)=6
• P2: 4(p1) =4
• P3: 4(p1)+3(p2)=7
 average waiting time =(6+4+7)/3 = 5.67ms
 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

16
Time Quantum and Context Switch Time

Performance
q large  FIFO (same FCFS)
q small  q must be large with respect to context
switch, otherwise overhead is too high,
=> performance decrease

Turnaround Time Varies With The Time Quantum

Rule:
80% of CPU bursts should
be shorter than q

17
Scheduling Algorithms

1. First-Come, First-Served Scheduling

2. Shortest-Job-First Scheduling
3. Round-Robin Scheduling
4. Priority Scheduling
5. Multilevel Queue Scheduling
6. Multilevel Feedback Queue Scheduling

Priority Scheduling

 A priority number (integer) is associated with each process

 The CPU is allocated to the process with the highest priority
(usually, smallest integer  highest priority)
 Two schemes:
• Preemptive
• Nonpreemptive
 Problem  Starvation – low priority processes may never execute
 Solution  Aging – as time progresses increase the priority of the
process
 Note: SJF is priority scheduling where priority is the inverse of
predicted next CPU burst time

18
Example of Priority Scheduling

ProcessA arri Burst TimeT Priority

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

 Priority scheduling Gantt Chart

 Average waiting time (0 + 1 + 6 + 16 + 18)/5 = 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

19
Scheduling Algorithms

1. First-Come, First-Served Scheduling

2. Shortest-Job-First Scheduling
3. Round-Robin Scheduling
4. Priority Scheduling
5. Multilevel Queue Scheduling
6. Multilevel Feedback Queue Scheduling

Multilevel Queue
 The ready queue consists of multiple queues
 Example:
• Priority scheduling, where each priority has its separate queue.
• Schedule the process in the highest-priority queue!

20
Multilevel Queue

 Prioritization based upon process type

Scheduling Algorithms

1. First-Come, First-Served Scheduling

2. Shortest-Job-First Scheduling
3. Round-Robin Scheduling
4. Priority Scheduling
5. Multilevel Queue Scheduling
6. Multilevel Feedback Queue Scheduling

21
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

 Three queues:
• Q0 – RR with time quantum 8 milliseconds
• Q1 – RR time quantum 16 milliseconds
• Q2 – FCFS
• Q1 has higher priority than both Q2 and Q3
(Q1>Q2>Q3)

 Scheduling:
• 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, the process is moved to queue Q1
• 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

22
ex

0 8 16 20 24 32 48 60 64 68
Q1 P1(36) P2(20) P3(12) P3(12)
8 P2(16)
Q2 P1(24) P1(20) P1(20) P1(20) P1(20)
16 P2(12) P2(12) P2(12) P2(0)
P3(4) P3(4) P3(4) P3(0)
Q3 P1(4) P1(4) P1(4) P1(0)

Expl

Q1 Q2 Q1 Q1 Q2 Q2 Q2 Q3

 Tại 0: P1 vào Q1, run với q=8, not done, move Q2. Q1: null
 Tại 8: Do trong Q1 null, nên P1 trong Q2 run. Q2: P1
 Tại 16: P2 vào Q1 vì Q1 có ưu tiên cao hơn Q2 nên P2 được run trước, P1 bị ngắt, P1 còn 20 (36-8-8),
P2 run với q=8 đến time 24 sẽ còn 12 (20-8). Q1:P2, Q2:P1. Nhưng
 Tại 20: P3 tới Q1 nhưng P2 chưa run xong. Lúc này P2, P3 cùng trong Q1 nên P2 đc run xong với q=8,
còn 12 (20-8) nên move Q2. lúc này Q1: P3, Q2: P1,P2
 Tại 24: P3 ở Q1, do Q1 có ưu tiên cao hơn Q2 nên P3 được run trước, sau q=8 thì ngắt, P3 còn 4 (12-
8). P3 move Q2, lúc này Q1: null, Q2: P1,2,3.
 Tại 32: Theo RR, P1 trong Q2 sẽ run trước với q=16, P1 còn 4 (20-16). P1 move to Q3. trong Q2: P2,3
 Tại 48: theo RR, P2 trong Q2 run vì được ưu tiên hơn Q3, sau 12m thì P2 done
 Tại 60: P2 done, P3 run sau 4m còn lại thì done tại 64. Trong Q2 null, đến Q3 run
 Tại 64: P1 trong Q3 được run sau 4m còn lại thì done tại 68
Operating System Concepts – 10th Edition 5a.46 Silberschatz, Galvin and Gagne ©2018

23
Chapter 2: Process Management

Advanced CPU Scheduling

Outline
 Thread Scheduling
 Multi-Processor Scheduling
 Real-Time CPU Scheduling
 Operating Systems Examples
 Algorithm Evaluation

24
Objectives
 Describe various CPU scheduling algorithms
 Assess CPU scheduling algorithms based on scheduling criteria
 Explain the issues related to multiprocessor and multicore scheduling
 Describe various real-time scheduling algorithms
 Describe the scheduling algorithms used in the Windows, Linux, and
Solaris operating systems
 Apply modeling and simulations to evaluate CPU scheduling
algorithms

Thread Scheduling
 Distinction between user-level and kernel-level threads
 When threads supported, threads scheduled, not processes
 Many-to-one and many-to-many models, thread library schedules
user-level threads to run on LWP (light weight process)
• Known as process-contention scope (PCS) since scheduling
competition is within the process
• Typically done via priority set by programmer
 Kernel thread scheduled onto available CPU is system-contention
scope (SCS) – competition among all threads in system

25
Thread Scheduling

 Lập lịch cục bộ (Local Scheduling):

• Bằng cách nào Thư viện luồng quyết định chọn luồng nào để đặt vào
một CPU ảo khả dụng:
 Thường chọn luồng có mức ưu tiên cao nhất
• Sự cạnh tranh CPU diễn ra giữa các luồng của cùng một tiến trình.
• Trong các HĐH sử dụng mô hình Many-to-one, Many-to-many.

 Lập lịch toàn cục (Global Scheduling)

• Bằng cách nào kernel quyết định kernel thread nào để lập lịch CPU
chạy tiếp.
• Sự cạnh tranh CPU diễn ra giữa tất cả các luồng trong hệ thống.
• Trong các HĐH sử dụng mô hình One-to-one (Windows XP, Linux,
Solaris 9)

Pthread Scheduling
 API allows specifying either PCS or SCS during thread creation
• PTHREAD_SCOPE_PROCESS schedules threads using PCS
scheduling
• PTHREAD_SCOPE_SYSTEM schedules threads using SCS
scheduling
 Can be limited by OS – Linux and macOS only allow
PTHREAD_SCOPE_SYSTEM

26
Pthread Scheduling API
#include <pthread.h>
#include <stdio.h>
#define NUM_THREADS 5
int main(int argc, char *argv[]) {
int i, scope;
pthread_t tid[NUM THREADS];
pthread_attr_t attr;
/* get the default attributes */
pthread_attr_init(&attr);
/* first inquire on the current scope */
if (pthread_attr_getscope(&attr, &scope) != 0)
fprintf(stderr, "Unable to get scheduling scope\n");
else {
if (scope == PTHREAD_SCOPE_PROCESS)
printf("PTHREAD_SCOPE_PROCESS");
else if (scope == PTHREAD_SCOPE_SYSTEM)
printf("PTHREAD_SCOPE_SYSTEM");
else
fprintf(stderr, "Illegal scope value.\n");
}

Pthread Scheduling API

/* set the scheduling algorithm to PCS or SCS */

pthread_attr_setscope(&attr, PTHREAD_SCOPE_SYSTEM);
/* create the threads */
for (i = 0; i < NUM_THREADS; i++)
pthread_create(&tid[i],&attr,runner,NULL);
/* now join on each thread */
for (i = 0; i < NUM_THREADS; i++)
pthread_join(tid[i], NULL);
}
/* Each thread will begin control in this function */
void *runner(void *param)
{
/* do some work ... */
pthread_exit(0);
}

27
Outline
 Thread Scheduling
 Multi-Processor Scheduling
 Real-Time CPU Scheduling
 Operating Systems Examples
 Algorithm Evaluation

Multiple-Processor Scheduling
 CPU scheduling more complex when multiple CPUs are available
 Multiprocess may be any one of the following architectures:
• Multicore CPUs
• Multithreaded cores
• NUMA systems
• Heterogeneous multiprocessing
• Homogeneous

28
Multiple-Processor Scheduling

 Symmetric multiprocessing (SMP) is where each processor is self

scheduling.
 All threads may be in a common ready queue (a)
 Each processor may have its own private queue of threads (b)

Multicore Processors
 Recent trend to place multiple processor cores on same physical chip
 Faster and consumes less power
 Multiple threads per core also growing
• Takes advantage of memory stall to make progress on another
thread while memory retrieve happens
 Figure

29
Multithreaded Multicore System
 Each core has > 1 hardware threads.
 If one thread has a memory stall, switch to another thread!
 Figure

Multithreaded Multicore System

 Chip-multithreading (CMT)
assigns each core multiple
hardware threads. (Intel refers to
this as hyperthreading.)

 On a quad-core system with 2

hardware threads per core, the
operating system sees 8 logical
processors.

30
Multithreaded Multicore System

 Two levels of scheduling:

1. The operating system

deciding which software
thread to run on a logical
CPU

2. How each core decides

which hardware thread to
run on the physical core.

Multiple-Processor Scheduling – Load Balancing

 If SMP (Symmetric multiprocessing), need to keep all CPUs loaded for

efficiency
 Load balancing attempts to keep workload evenly distributed
 Push migration – periodic task checks load on each processor, and if
found pushes task from overloaded CPU to other CPUs
 Pull migration – idle processors pulls waiting task from busy
processor

31
Multiple-Processor Scheduling – Processor Affinity

 When a thread has been running on one processor, the cache contents
of that processor stores the memory accesses by that thread.
• We refer to this as a thread having affinity for a processor (i.e.,
“processor affinity”)
 Load balancing may affect processor affinity as a thread may be moved
from one processor to another to balance loads, yet that thread loses
the contents of what it had in the cache of the processor it was moved
off of.
 Soft affinity – the operating system attempts to keep a thread running
on the same processor, but no guarantees.
 Hard affinity – allows a process to specify a set of processors it may
run on.

NUMA and CPU Scheduling

• NUMA (non-uniform memory access), is a method of configuring a cluster of
microprocessor in a multiprocessing system so that they can share memory
locally, improving performance and the ability of the system to be expanded.
• NUMA is used in a symmetric multiprocessing ( SMP ) system.
• If the operating system is NUMA-aware, it will assign memory closes to the
CPU the thread is running on.

32
Outline
 Thread Scheduling
 Multi-Processor Scheduling
 Real-Time CPU Scheduling
 Operating Systems Examples
 Algorithm Evaluation

Real-Time CPU Scheduling

 Can present obvious challenges
 Hard real-time systems –
• task must be serviced by its deadline
 Soft real-time systems –
• Critical real-time tasks have the highest priority, but no guarantee
as to when tasks will be scheduled

33
Real-Time CPU Scheduling
 Event latency – the amount of time
that elapses from when an event
occurs to when it is serviced.
 Two types of latencies affect
performance
1. Interrupt latency – time from
arrival of interrupt to start of
routine that services interrupt
2. Dispatch latency – time for
schedule to take current process
off CPU and switch to another

Interrupt Latency
The interrupt latency refers to the delay between the start of an Interrupt
Request (IRQ) and the start of the respective Interrupt Service Routine (ISR).
The interrupt latency is expressed in core clock cycles.

34
Dispatch Latency

 Conflict phase of dispatch

latency:
1. Preemption of any process
running in kernel mode
2. Release by low-priority
process of resources needed
by high-priority processes

Priority-based Scheduling
 For real-time scheduling, scheduler must support preemptive, ex:
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

35
Rate Monotonic Scheduling
 A priority is assigned based on the inverse of its period
 Shorter periods = higher priority;
 Longer periods = lower priority
 P1 is assigned a higher priority than P2.

Missed Deadlines with Rate Monotonic Scheduling

 Process P2 misses finishing its deadline at time 80

 Figure

36
Earliest Deadline First Scheduling (EDF)

 Priorities are assigned according to deadlines:

• The earlier the deadline, the higher the priority
• The later the deadline, the lower the priority
 Figure

Proportional Share Scheduling

 T shares are allocated among all processes in the system

 An application receives N shares where N < T
 This ensures each application will receive N / T of the total
processor time

37
POSIX Real-Time Scheduling
 The POSIX.1b standard
 API provides functions for managing real-time threads
 Defines two scheduling classes for real-time threads:
1. SCHED_FIFO - threads are scheduled using a FCFS strategy with
a FIFO queue. There is no time-slicing for threads of equal priority
2. SCHED_RR - similar to SCHED_FIFO except time-slicing occurs
for threads of equal priority
 Defines two functions for getting and setting scheduling policy:
1. pthread_attr_getsched_policy(pthread_attr_t
*attr, int *policy)
2. pthread_attr_setsched_policy(pthread_attr_t
*attr, int policy)

POSIX Real-Time Scheduling API

#include <pthread.h>
#include <stdio.h>
#define NUM_THREADS 5
int main(int argc, char *argv[])
{
int i, policy;
pthread_t_tid[NUM_THREADS];
pthread_attr_t attr;
/* get the default attributes */
pthread_attr_init(&attr);
/* get the current scheduling policy */
if (pthread_attr_getschedpolicy(&attr, &policy) != 0)
fprintf(stderr, "Unable to get policy.\n");
else {
if (policy == SCHED_OTHER) printf("SCHED_OTHER\n");
else if (policy == SCHED_RR) printf("SCHED_RR\n");
else if (policy == SCHED_FIFO) printf("SCHED_FIFO\n");
}

38
POSIX Real-Time Scheduling API (Cont.)

/* set the scheduling policy - FIFO, RR, or OTHER */

if (pthread_attr_setschedpolicy(&attr, SCHED_FIFO) != 0)
fprintf(stderr, "Unable to set policy.\n");
/* create the threads */
for (i = 0; i < NUM_THREADS; i++)
pthread_create(&tid[i],&attr,runner,NULL);
/* now join on each thread */
for (i = 0; i < NUM_THREADS; i++)
pthread_join(tid[i], NULL);
}

/* Each thread will begin control in this function */

void *runner(void *param)
{
/* do some work ... */
pthread_exit(0);
}

Outline
 Thread Scheduling
 Multi-Processor Scheduling
 Real-Time CPU Scheduling
 Operating Systems Examples
 Algorithm Evaluation

39
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

40
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 task with lowest virtual
run time

41
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

42
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

43
Windows Priority Classes
 Win32 API identifies several priority classes to which a process can
belong
• REALTIME_PRIORITY_CLASS, HIGH_PRIORITY_CLASS,
ABOVE_NORMAL_PRIORITY_CLASS,NORMAL_PRIORITY_CL
ASS, BELOW_NORMAL_PRIORITY_CLASS,
IDLE_PRIORITY_CLASS
• All are variable except REALTIME
 A thread within a given priority class has a relative priority
• TIME_CRITICAL, HIGHEST, ABOVE_NORMAL, NORMAL,
BELOW_NORMAL, LOWEST, IDLE
 Priority class and relative priority combine to give numeric priority
 Base priority is NORMAL within the class
 If quantum expires, priority lowered, but never below base

Windows Priority Classes (Cont.)

 If wait occurs, priority boosted depending on what was waited for
 Foreground window given 3x priority boost
 Windows 7 added user-mode scheduling (UMS)
• Applications create and manage threads independent of kernel
• For large number of threads, much more efficient
• UMS schedulers come from programming language libraries like
C++ Concurrent Runtime (ConcRT) framework

44
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

45
Solaris Dispatch Table

Solaris Scheduling

46
Solaris Scheduling (Cont.)
 Scheduler converts class-specific priorities into a per-thread
global priority
• Thread with highest priority runs next
• Runs until (1) blocks, (2) uses time slice, (3) preempted by
higher-priority thread
• Multiple threads at same priority selected via RR

Outline
 Thread Scheduling
 Multi-Processor Scheduling
 Real-Time CPU Scheduling
 Operating Systems Examples
 Algorithm Evaluation

47
Algorithm Evaluation
 How to select CPU-scheduling algorithm for an OS?
 Determine criteria, then evaluate algorithms
 Deterministic modeling
• Type of analytic evaluation
• Takes a particular predetermined workload and defines
the performance of each algorithm for that workload
 Consider 5 processes arriving at time 0:

Deterministic Evaluation

 For each algorithm, calculate minimum average waiting time

 Simple and fast, but requires exact numbers for input,
applies only to those inputs
• FCS is 28ms:

• Non-preemptive SFJ is 13ms:

• RR is 23ms (q=10):

48
Queueing Models
 Describes the arrival of processes, and CPU and I/O bursts
probabilistically
• Commonly exponential, and described by mean
• Computes average throughput, utilization, waiting time, etc.
 Computer system described as network of servers, each with queue of
waiting processes
• Knowing arrival rates and service rates
• Computes utilization, average queue length, average wait time,
etc.

Little’s Formula
 n = average queue length
 W = average waiting time in queue
 λ = average arrival rate into queue
 Little’s law – in steady state, processes leaving queue must
equal processes arriving, thus:
n=λxW
• Valid for any scheduling algorithm and arrival distribution
 For example, if on average 7 processes arrive per second, and
normally 14 processes in queue, then average wait time per
process = 2 seconds

49
Simulations
 Queueing models limited
 Simulations more accurate
• Programmed model of computer system
• Clock is a variable
• Gather statistics indicating algorithm performance
• Data to drive simulation gathered via
 Random number generator according to probabilities
 Distributions defined mathematically or empirically
 Trace tapes record sequences of real events in real systems

Evaluation of CPU Schedulers by Simulation

50
Implementation
 Even simulations have limited accuracy
 Just implement new scheduler and test in real systems
• High cost, high risk
• Environments vary
 Most flexible schedulers can be modified per-site or per-system
 Or APIs to modify priorities
 But again environments vary

End of Chapter 5

TcIC V13.2.1.2 - Installation Manual
No ratings yet
TcIC V13.2.1.2 - Installation Manual
6 pages
Operating Systems Lecture Notes-1
100% (1)
Operating Systems Lecture Notes-1
15 pages
Chapter 5: CPU Scheduling: Silberschatz, Galvin and Gagne ©2018 Operating System Concepts - 10 Edition
No ratings yet
Chapter 5: CPU Scheduling: Silberschatz, Galvin and Gagne ©2018 Operating System Concepts - 10 Edition
43 pages
10 ch5 up
No ratings yet
10 ch5 up
58 pages
OS_CH5 (2)
No ratings yet
OS_CH5 (2)
115 pages
Ch5 en CPUSched 2024
No ratings yet
Ch5 en CPUSched 2024
62 pages
CPU Scheduling
No ratings yet
CPU Scheduling
81 pages
Lec6 - Ch5- CPU Scheduling
No ratings yet
Lec6 - Ch5- CPU Scheduling
32 pages
Chap 4 CPU Scheduling
No ratings yet
Chap 4 CPU Scheduling
21 pages
ch5 - CPU Scheduling UPDATED WITH NOTES SHORT
No ratings yet
ch5 - CPU Scheduling UPDATED WITH NOTES SHORT
42 pages
ch5 Part1
No ratings yet
ch5 Part1
12 pages
Chapter 5a - CPU Scheduling
No ratings yet
Chapter 5a - CPU Scheduling
20 pages
Chapter 5 CPU Scheduling
No ratings yet
Chapter 5 CPU Scheduling
58 pages
CH 5 Process Scheduling STD
No ratings yet
CH 5 Process Scheduling STD
100 pages
Ch05 (CPU Scheduling)
No ratings yet
Ch05 (CPU Scheduling)
33 pages
Chapter-5 (9) - Copy
No ratings yet
Chapter-5 (9) - Copy
69 pages
Ch5 Updated
No ratings yet
Ch5 Updated
61 pages
ch5a
No ratings yet
ch5a
30 pages
Ch5 Revised
No ratings yet
Ch5 Revised
41 pages
ch5 updated- operating system
No ratings yet
ch5 updated- operating system
45 pages
CH 5
No ratings yet
CH 5
74 pages
OS-ch4-Process Scheduling
No ratings yet
OS-ch4-Process Scheduling
32 pages
Chapter 5 - CPU (Process) Scheduling
No ratings yet
Chapter 5 - CPU (Process) Scheduling
42 pages
Chapter 5a: CPU Scheduling: Silberschatz, Galvin and Gagne ©2013 Operating System Concepts - 9 Edition
No ratings yet
Chapter 5a: CPU Scheduling: Silberschatz, Galvin and Gagne ©2013 Operating System Concepts - 9 Edition
33 pages
Chapter 5: CPU Scheduling: Silberschatz, Galvin and Gagne ©2018 Operating System Concepts
No ratings yet
Chapter 5: CPU Scheduling: Silberschatz, Galvin and Gagne ©2018 Operating System Concepts
76 pages
Unit-4 CPU Scheduling
No ratings yet
Unit-4 CPU Scheduling
8 pages
(502047) NhapmonHedieuhanh Chuong5
No ratings yet
(502047) NhapmonHedieuhanh Chuong5
80 pages
ch5 OS
No ratings yet
ch5 OS
26 pages
ch5.pptx
No ratings yet
ch5.pptx
79 pages
CH 5
No ratings yet
CH 5
80 pages
Unit II: Process Management: Dr.S.Kavi Priya, ASP/CSE, MSEC 9842295563
No ratings yet
Unit II: Process Management: Dr.S.Kavi Priya, ASP/CSE, MSEC 9842295563
201 pages
ch5_a
No ratings yet
ch5_a
68 pages
CH 5
No ratings yet
CH 5
44 pages
CH 5 Process Scheduling
No ratings yet
CH 5 Process Scheduling
100 pages
ch5 CPU Scheduling
No ratings yet
ch5 CPU Scheduling
72 pages
Chapter 4 CPU Scheduling
No ratings yet
Chapter 4 CPU Scheduling
65 pages
Ch5 - Cpu Scheduling Updated With Notes
No ratings yet
Ch5 - Cpu Scheduling Updated With Notes
49 pages
ch6 CPUSheduling
No ratings yet
ch6 CPUSheduling
74 pages
Unit 2 (CPU Scheduling)
No ratings yet
Unit 2 (CPU Scheduling)
27 pages
ch5b
No ratings yet
ch5b
46 pages
Lecture 5 CPU Scheduling
No ratings yet
Lecture 5 CPU Scheduling
40 pages
Unit 2
No ratings yet
Unit 2
120 pages
Week 4b
No ratings yet
Week 4b
13 pages
Operating System - Process Scheduling
No ratings yet
Operating System - Process Scheduling
40 pages
CH 05
No ratings yet
CH 05
83 pages
ch6 1
No ratings yet
ch6 1
63 pages
CH 6
No ratings yet
CH 6
78 pages
OS Lec 4
No ratings yet
OS Lec 4
19 pages
chapter 5
No ratings yet
chapter 5
53 pages
CSE 315 Lec-03
No ratings yet
CSE 315 Lec-03
43 pages
Chapter 5: CPU Scheduling: Silberschatz, Galvin and Gagne ©2009 Operating System Concepts - 8 Edition
No ratings yet
Chapter 5: CPU Scheduling: Silberschatz, Galvin and Gagne ©2009 Operating System Concepts - 8 Edition
46 pages
Unit 04
No ratings yet
Unit 04
22 pages
Ch5 CPU Scheduling
No ratings yet
Ch5 CPU Scheduling
38 pages
ICS 431-Ch5-Processes Scheduling
No ratings yet
ICS 431-Ch5-Processes Scheduling
129 pages
Lecture 4 - CPU Scheduling
No ratings yet
Lecture 4 - CPU Scheduling
26 pages
05 - CPU Scheduling
No ratings yet
05 - CPU Scheduling
15 pages
CH 3
No ratings yet
CH 3
29 pages
Lec 3 & 4
No ratings yet
Lec 3 & 4
57 pages
Tutorial QNS On Process MGMT - 032041
No ratings yet
Tutorial QNS On Process MGMT - 032041
6 pages
Chapter 4 Notes
No ratings yet
Chapter 4 Notes
34 pages
Module2 RTOS
No ratings yet
Module2 RTOS
56 pages
Advanced Backend Code Optimization
From Everand
Advanced Backend Code Optimization
Sid Touati
No ratings yet
N25 B3 C2
No ratings yet
N25 B3 C2
7 pages
Bai Thi KTMT k15 3 Trac Nghiem Kien Truc May Tinh
No ratings yet
Bai Thi KTMT k15 3 Trac Nghiem Kien Truc May Tinh
19 pages
Thi Kien Truc May Tinh Va Hop Ngu
No ratings yet
Thi Kien Truc May Tinh Va Hop Ngu
15 pages
ch02.5 - Explain - Excercise - Deadlock
No ratings yet
ch02.5 - Explain - Excercise - Deadlock
12 pages
ch02.4 - Explain - Excercise - Synch
No ratings yet
ch02.4 - Explain - Excercise - Synch
5 pages
ch02.5 - Deadlock
No ratings yet
ch02.5 - Deadlock
21 pages
ch02.12. SocketMultithread in Python
No ratings yet
ch02.12. SocketMultithread in Python
22 pages
ch02.1 - Process
No ratings yet
ch02.1 - Process
35 pages
Lock-Based Protocols
No ratings yet
Lock-Based Protocols
8 pages
Lecture Notes 09
No ratings yet
Lecture Notes 09
29 pages
CH 2
No ratings yet
CH 2
50 pages
Computer Architecture Assignment 3 (ARCH)
No ratings yet
Computer Architecture Assignment 3 (ARCH)
9 pages
Storage Lab Exam Guide
No ratings yet
Storage Lab Exam Guide
10 pages
Installing Oracle 9i On RHEL5
No ratings yet
Installing Oracle 9i On RHEL5
7 pages
Logcat
No ratings yet
Logcat
26 pages
Network Virtualization For FreeBSD Jails Lecture Note
No ratings yet
Network Virtualization For FreeBSD Jails Lecture Note
2 pages
Principles of Operating Systems: Lecture 5 - Interprocess Communication Ardalan Amiri Sani
No ratings yet
Principles of Operating Systems: Lecture 5 - Interprocess Communication Ardalan Amiri Sani
38 pages
Command-Line Support Tools
No ratings yet
Command-Line Support Tools
1 page
Red Hat Enterprise Linux-7-Performance Tuning Guide-en-US PDF
No ratings yet
Red Hat Enterprise Linux-7-Performance Tuning Guide-en-US PDF
115 pages
How To Solve Cannot Installed To Hadisk GPT Partition
No ratings yet
How To Solve Cannot Installed To Hadisk GPT Partition
5 pages
Linux Os-1
No ratings yet
Linux Os-1
56 pages
Log
No ratings yet
Log
7 pages
GTX 1070 Vbios 86.04.50 Update Guide v1.0
No ratings yet
GTX 1070 Vbios 86.04.50 Update Guide v1.0
5 pages
Lecture 13
No ratings yet
Lecture 13
53 pages
Introduction To UNIX
No ratings yet
Introduction To UNIX
53 pages
CyberArk Issues
No ratings yet
CyberArk Issues
3 pages
Deep Spar Disk Imager
No ratings yet
Deep Spar Disk Imager
3 pages
CH18 COA9e
No ratings yet
CH18 COA9e
28 pages
KSH 93 Quick Ref Builtins
No ratings yet
KSH 93 Quick Ref Builtins
2 pages
Operating System
No ratings yet
Operating System
2 pages
ECRIN v4.20 - App-V Sequencing Recipe
No ratings yet
ECRIN v4.20 - App-V Sequencing Recipe
12 pages
CICS
100% (3)
CICS
109 pages
February Month Attendence Engineer 2
No ratings yet
February Month Attendence Engineer 2
1 page
Linux
No ratings yet
Linux
397 pages
Osintgram : Osintgram Is A OSINT Tool On Instagram To Collect, Analyze, and Run Reconnaissance
No ratings yet
Osintgram : Osintgram Is A OSINT Tool On Instagram To Collect, Analyze, and Run Reconnaissance
5 pages
Download Modern Operating Systems 5th Edition Andrew S. Tanenbaum ebook All Chapters PDF
100% (2)
Download Modern Operating Systems 5th Edition Andrew S. Tanenbaum ebook All Chapters PDF
51 pages