Day 25Uniprocessor scheduling

Algorithm: Selection function

Selection function – which process among ready processes to select.

w – time spent in system, waiting and executing, so far

e – time spent in execution, so far s- total service time required by the process

Service time – the time that the process actually runs (includes e)

Algorithm: Decision Mode Non-pre-emptive

Once a process is in the running state, it will continue until it terminates or blocks itself for I/O

Selection function is used only when a process blocks or terminates

Pre-emptive Currently running process may be forced to release the

processor and moved to the Ready state by the operating system

Allows for better service since any one process cannot monopolize the processor for very long

More switching overhead Selection function is used when a process blocks or terminates

or when a new process enters the system or when a time slice expires.

First-Come-First-Served(FCFS/FIFO)

Each process joins the Ready queue When the current process ceases to execute,

the oldest process in the Ready queue is selected non-pre-emptive

A short process may have to wait a very long time before it can execute

FCFS

Arrival time Service Time

P1 0 24

P2 1 3

P3 2 3

Calculate the turn-around time for each process and the average turn-around time.

FCFS

Favors CPU-bound processes I/O processes have to wait until CPU-bound

process completes Inefficient use of processor and I/O devices. Rarely used on its own.

Process Scheduling Example

First-Come-First-Served(FCFS)

0 5 10 15 20

A

B

C

D

E

Shortest Process Next (SPN)

Select the shortest process in the ready queue.

Non-pre-emptive. Reduces the response time for shorter

processes. Less predictable response time for longer

processes. Must estimate the running time for each

process.

Process Scheduling Example

Shortest Process Next (SPN)

0 5 10 15 20

A

B

C

D

E

Shortest Remaining Time(SRT)

Pre-emptive version of SPN. Check when a new process enters the

system and when the current process voluntarily releases the processor.

Better response time as short jobs given immediate preference.

Fewer process switches than RR. Must keep track of elapsed time.

Process Scheduling Example

Shortest Remaining Time(SRT)

0 5 10 15 20

A

B

C

D

E

Round robin

Pre-empt a process using the clock interrupt. A process is allowed to run for a time-slice or

quantum. The next process in the Ready queue is

selected to run following the pre-emption (FCFS).

Prevents large processes from monopolizing the processor.

Round robin

Design issue: length of quantum Short quantum

short processes will finish quickly. lots of process switches

Choose such that typical processes can complete without being interrupted.

Round robin

Unfair to I/O-bound processes as they will typically not use a complete quantum, while CPU-bound processes will.

Use Virtual Round Robin Use an auxiliary queue to hold processes that

have I/O requests serviced. Process in auxiliary queue gets preference over

those in the main ready queue. Process from the auxiliary runs for time remaining

from last time-slice.

Process Scheduling Example

Round-Robin

0 5 10 15 20

A

B

C

D

E

Highest Response Ratio Next (HRRN)

ss w

ratio Response

w – time spent waiting for the processor

s – expected service time

HRRN

Non-pre-emptive Select the process that has the highest

response ratio. Short process (small “s”) given preference. Long process(large “s”) that has been waiting

a long time (large “w”) is given preference. Accounts for age of process.

Must estimate “s”.

Process Scheduling Example

Highest Response Ratio Next (HRRN)

A

B

C

D

E

0 5 10 15 20

Multi-level Feedback (quantum = i)

Pre-emptive (time quantum) Maintain many ready queues of varying priority.

Each queue uses FCFS. A process that is pre-empted is moved to a queue of

lower priority to wait. Thus longer processes are penalized. Don’t have to estimate running time. Note: If there are no processes waiting, don’t pre-

empt the process

Process Scheduling Example

Multi-level Feedback (quantum = i)

0 5 10 15 20

A

B

C

D

E

Multi-level Feedback (quantum = 2i * quantum)

With quantum = i, long processes will have very high turn-around time.

So, a process in queue RQi will be allowed to run for 2i * quantum.

Multi-level Feedback (quantum = 2i * quantum)

Fair-Share Scheduling

Treat multiple processes and threads of a user as a collection.

The performance of the collection of threads is optimized not just that of any one thread or process.

Fair share scheduling ensures that users are given a fair share of the processor time.

FSS

Divide user community into groups. Each user process’ priority is based on the

process’ execution time and the execution time of processes within the same group.