OS 4 CPUScheduling

Operating Systems
ECEG-5202
CPU Scheduling
Surafel Lemma Abebe (Ph. D.)

Outline
• Basic Concepts
• Scheduling Criteria
• Process Scheduling
• Multiple-Processor Scheduling
Surafel Lemma Abebe (Ph. D.) 2

Basic concepts
• At a time, on a single processor only one process
can run
• What is the objective of multi-programming?
– Maximize CPU utilization by having some process
running at all times
• Process execution consist of
– A cycle of CPU execution
• Starts at CPU burst
– A cycle of I/O wait
• Starts at I/O burst

Basic concepts…
• CPU-I/O burst cycle
– Process execution alternates between
• CPU burst cycle
• I/O burst cycle
– I/O bound program has many short CPU
bursts
– CPU bound program has a few long CPU
bursts
– Best performance could be achieved by
combining I/O-bound and CPU-bound process

Basic concepts…
 Schedulers
 Short-term scheduler (or CPU scheduler)
 Selects which process should be executed next and allocates CPU
 Sometimes the only scheduler in a system
 Short-term scheduler is invoked frequently (in milliseconds)  (must be fast)
 Long-term scheduler (or job scheduler)
 Selects which processes (from those spooled to a mass storage, e.g., in a batch system)
should be brought into the ready queue
 Long-term scheduler is invoked infrequently (in seconds, minutes)  (may be slow)
 The long-term scheduler controls the degree of multiprogramming (# of processes in
memory)
 Absent in time sharing systems such as UNIX and MS Windows
 Medium-term scheduler
 Can be added if degree of multiple programming needs to be decreased by
removing a process from memory
 Remove process from memory, store on disk, bring back in from disk to
continue execution: swapping

Basic concepts…
• CPU scheduler
– Also called short-term scheduler
– Carries out the selection of a process from the
processes in memory that are ready to execute
– Allocates the CPU to processes that are in memory
and ready to execute
– Ready queue could be ordered in various ways
– Records in the queue are generally process control
blocks (PCBs) of the processes

Basic concepts…
• Dispatcher
– Component involved in the CPU scheduling function
– Gives control of the CPU to the process selected by the
short-term scheduler
– This function involves the following
• Switching context
• Switching to user mode
• Jumping to the proper location in the user program to restart that
program
– Should be as fast as possible
• Invoked during every process switch
– Dispatch latency
• Time it takes for the dispatcher to stop one process and start
another running

Scheduling criteria
• Used to compare CPU-scheduling algorithms
• The criteria affects the judgment of the best
algorithm
• Properties of algorithms
– CPU utilization
• Refers to a computer's usage of CPU, or the amount of work
handled by a CPU
• Percent of the CPU’s cycles being spent on your process
– Throughput
• Refers to number of processes that are completed per time
unit
• One completed process per hour vs ten completed
processes per second

Scheduling criteria…
– Turnaround time
• Refers to the time interval from submission of a process to
the time of completion
– This is from the point of view of a particular process
• Is the sum of the time spent waiting to get into memory,
waiting in the ready queue, executing on the CPU, and doing
I/O
– Waiting time
• Refers to the sum of the periods a process spends waiting in
ready queue
– Response time
• Refers to the time from the submission of a request until the
start of the first response
– Not the time it takes to output the response
• Limited by the speed of the output device
Scheduling criteria…
• Ideally, choose a CPU scheduler that optimizes all criteria
simultaneously
– Maximize CPU utilization and throughput
– Minimize turnaround time, waiting time, and response
time
=> Generally not possible
• Choose a scheduling algorithm based on its ability to satisfy
a policy
– Minimize average response time
• Provide output to the user as quickly as possible
• Process their input as soon as it is received
– Minimize variance of response time
• In interactive systems, predictability may be more important
than a faster average response time
Process Scheduling
• Goals of scheduling
– Fairness
• Each process gets its fair share of the CPU and no process can suffer indefinite
postponement
• Giving equivalent or equal time is not always fair
– E.g., safety control and payroll at a nuclear plant
– Policy enforcement
• The systems policy must be enforced
– Efficiency
• CPU should be kept busy for almost 100% of the time
– Response time
• Response time for interactive users must be minimized
– Turnaround
• The time batch users must wait for an output should be minimized
– Throughput
• The number of jobs processed per unit time should be maximized

Process Scheduling…
• Maintains scheduling queues of processes
• Types of queues
– Job queue – set of all processes in the system
– Ready queue – set of all processes residing in main memory, ready and waiting to
execute
– Device queues – set of processes waiting for an I/O device
• Processes could move among the various queues

• Queuing diagram
– Represents queues, resources, flows

Process Scheduling …
• Scheduling algorithms
– Can be divided into two categories based on
how they deal with clock interrupts
– Non-preemptive scheduling
• Once a process has been given the CPU, the CPU
cannot be taken away from that process
• Characteristics
– Short jobs are made to wait by longer jobs but the overall
treatment of all processes is fair
– Response times are more predictable because incoming high
priority jobs can not displace waiting jobs

• Scheduling algorithms…
– Preemptive scheduling
• Once a process has been given CPU, the CPU can be taken away
• 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
• In case of 1 and 4, the scheduling is non-preemptive or cooperative
– Mostly used in older versions of OS
– Cooperative scheduling
• Doesn’t require the special hardware (e.g., timer) needed for preemptive
scheduling
• Scheduling for 2 and 3 is preemptive
• Could create race condition. How?
– First come first served
• Also called First-In-First-Out (FIFO), Run-to-Completion, Run-until-
Done
• Simplest scheduling algorithm
• Processes are dispatched according to their arrival time on the ready
queue
• Once process has a CPU, it runs to completion
• Advantage
– Fair
– More predictable
– Simple to write and understandable code
• Disadvantage
– Long jobs make short jobs wait
– Unimportant jobs make important jobs wait
– Not useful in scheduling interactive users, because it cannot guarantee good
response time
– Average time is often quite long

– Shortest-Job-First (SJF)
• Waiting process with the smallest estimated run-time-to-
completion is run next
• Appropriate for batch jobs for which the run times are known in
advance
• Favors short jobs (or processes) at the expense of longer ones
• Could be preemptive or non-preemptive
– Preemptive occurs when a new process arrives in the ready queue that
has a predicted burst time shorter than the time remaining in the process
whose burst is currently on the CPU
• Advantage
– Gives the minimum average time for a given set of processes
• Disadvantage
– Requires precise knowledge of how long a job or process will run
» Relies on user estimates of run times

– Priority
• Each process is assigned a priority and allowed to run according to its
priority
• Equal-priority processes are scheduled in FCFS order
• Could be defined internally or externally
– Internal priorities
» Time limits, memory requirement, file requirement, CPU vs I/O
requirements
– External defined priorities are set by criteria that are external to OS
» Importance of process, amount of funds being paid for computer use, politics
• Can be either preemptive or non-preemptive
– Preemptive
» When process with higher priority arrives
• Problem
– Indefinite blocking or starvation of lower priority processes
• Solution
– Aging
» Gradually increases the priority of processes that wait in the system for a
longer period
– Round robin
• Processes are dispatched in a FIFO manner but are given a limited
amount of CPU time called a time-slice or a quantum
• The oldest, simplest, fair and most widely used
• Preemptive
– Effective in time-sharing environment
– Guarantees reasonable response time for interactive users
• Challenge
– Setting the length of the quantum
» Too short quantum
• Causes too many context switches and lower the CPU efficiency
» Too long quantum
• Causes poor response time
• Average waiting time is often quite long

– Multilevel queue
• Partitions the ready queue in several separate queues
• Processes are assigned to a queue permanently based on
– Memory size, process priority, process type, …
• Each queue has its own scheduling algorithm policy
• Could use either preemptive or non-preemptive
• Possibility I
– If each queue has absolute priority over lower priority queues, wait for all processes in
highest priority queue finishes
• Possibility II
– If there is a time slice between the queues then each queue gets a certain amount of
CPU time, which it can then schedule among the processes in its queue
– Example:
» 80% of the CPU time to foreground queue using RR
» 20% of the CPU time to background queue using FCFS
• Has advantage of low scheduling overhead but its inflexible

– Multilevel feedback queue
• Same as multilevel queue but allows a process to move between
queues
– If the characteristics of a job changes between CPU-intensive and I/O
intensive
– If a process waits too long in the lower-priority queue, it may be moved
to a higher priority queue
• Is the most flexible as it can be tuned for any situation
• Complex to implement because of all the adjustable parameters
– The number of queues
– The scheduling algorithm for each queue
– The methods used to upgrade or demote
processes from one queue to another
– The method used to determine which queue
a process enters initially

Multi-processor scheduling
• Multiple CPUs
=> load sharing
– But scheduling becomes more complex
• Approaches
– Asymmetric multiprocessing
• All scheduling decisions, I/O processing, and other system
activities are handled by a single processor
– Master server
• Other processors execute only user code
• Simple
– One processor accesses the system data structure, reducing the
need for data sharing

Multi-processor scheduling…
• Approaches..
– Symmetric multiprocessing (SMP)
• Each processor is self-scheduling
• Each processor could use its own ready queue or a
common ready queue
• Care must be taken to ensure
– Two processors do not choose to schedule the same process
– Processes are not lost from the queue
• Supported by many modern operating systems

• Process Affinity
– Keeps a process closer to the processor it has first
executed
– Avoids the high cost of invalidating and repopulating
caches
– Soft affinity
• Follows a policy of attempting to keep a process running on
the same processor but its not guaranteed
– Hard affinity
• Allows a process to specify a subset of processors on which
it may run

• Process Affinity…
– Main-memory architecture of a system can affect processor
affinity issues
– NUMA (Non-uniform memory access)
• CPU has faster access to some parts of main memory than to other
parts
• Occurs in systems containing combined CPU and memory boards
• If a process is assigned to a CPU with closer affinity, it can be allocated
memory on the board where the CPU resides
– Requires the CPU scheduler and memory-placement algorithm to work
together

• Load balancing
– Attempts to keep the workload evenly distributed
across all processors in an SMP system
– Important only on systems where each processor has
its own private queue of eligible processes to execute
– Approaches
• Push migration
– A specific task periodically checks the load on each processor and
make sure that the load is balanced
• Pull migration
– An idle processor pulls a waiting task from a busy processor
– Load balancing sometimes contradicts with processor
affinity. Explain?
• Multicore processors
– Processors with multiple processor cores on the same physical chip
– To the OS, each core appears to be a separate physical processor
– Advantage over systems in which each processor has its own physical
chip
• They are faster
• They consume less power
– Problem
• Memory stall
– Memory stall
• The significant amount of time spent by a processor while waiting for a data to
be available
• Happens when a processor accesses memory (due to cache miss)

• Multicore processors…
– Solution
• Multithreaded processor cores
– Two or more hardware threads are assigned to each core
– If one thread stalls while waiting for memory, the core can switch
to another thread
– From OS perspective, each hardware thread appears as a logical
processor that is available to run a software thread
» In dual core, dual threaded system,
• There are four logical processors to the OS

• Multicore processors…
– Multithreaded processor cores…
• Require two levels of scheduling
– Which software thread to run on each logical processor
» OS uses any of the scheduling algorithms discussed
earlier
– How each core decides which hardware thread to run
» Several strategies
• Use a simple round robin algorithm
• Assign to each hardware thread a dynamic urgency
value ranging from 0 to 7 and pick the one with
highest urgency

OS 4 CPUScheduling

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

OS 4 CPUScheduling

Uploaded by

Copyright:

Available Formats

Operating Systems

Surafel Lemma Abebe (Ph. D.)

Surafel Lemma Abebe (Ph. D.) 2

Surafel Lemma Abebe (Ph. D.) 3

Surafel Lemma Abebe (Ph. D.) 4

Surafel Lemma Abebe (Ph. D.) 5

Surafel Lemma Abebe (Ph. D.) 6

Surafel Lemma Abebe (Ph. D.) 7

Surafel Lemma Abebe (Ph. D.) 8

Surafel Lemma Abebe (Ph. D.) 13

Surafel Lemma Abebe (Ph. D.) 14

Surafel Lemma Abebe (Ph. D.) 15

Surafel Lemma Abebe (Ph. D.) 16

• CPU scheduling decisions may take place when a process

Surafel Lemma Abebe (Ph. D.) 18

Surafel Lemma Abebe (Ph. D.) 19

Surafel Lemma Abebe (Ph. D.) 21

Surafel Lemma Abebe (Ph. D.) 22

Surafel Lemma Abebe (Ph. D.) 23

Surafel Lemma Abebe (Ph. D.) 24

Surafel Lemma Abebe (Ph. D.) 25

Surafel Lemma Abebe (Ph. D.) 26

Surafel Lemma Abebe (Ph. D.) 27

Surafel Lemma Abebe (Ph. D.) 29

Surafel Lemma Abebe (Ph. D.) 30

Surafel Lemma Abebe (Ph. D.) 31

You might also like