CPU SchedulingBasic ConceptsScheduling Criteria Scheduling AlgorithmsMultiple-Processor SchedulingAlgorithm Evaluation

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 distribution

Basic Concepts

CPU And I/O Bursts

Histogram of CPU-burst Times

Selects from among the processes in memory that are ready to execute, and allocates the CPU to one of them.

CPU scheduling decisions may take place when a process:◦ Switches from running to waiting state.◦ Switches from running to ready state.◦ Switches from waiting to ready.◦ Terminates.

Scheduling under 1 and 4 is nonpreemptive.

All other scheduling is preemptive.

CPU Scheduler

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.

Dispatcher

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

Scheduling Criteria

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 Criteria

Max CPU utilization Max throughput Min turnaround time Min waiting time Min response time

Optimization Criteria

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 schedule is:

Waiting time for P1 = 0; P2 = 24; P3 = 27 Average waiting time: (0 + 24 + 27)/3 = 17

First-Come, First-Served (FCFS)

P1 P2 P3

24 27 300

Suppose that the processes arrive in the order P2 , P3 , P1

The Gantt chart for the schedule is:

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

FCFS Scheduling (Cont.)

P1P3P2

63 300

Associate with each process the length of its next CPU burst. Use these lengths to schedule the process with the shortest time.

Two schemes: ◦ nonpreemptive – once CPU given to the process it

cannot be preempted until completes its CPU burst.

Shortest-Job-First (SJF) Scheduling

Preemptive Version (SRTF)◦ if a new process arrives with CPU burst length less

than remaining time of current executing process, preempt. This scheme is know as the Shortest-Remaining-Time-First (SRTF).

SJF is optimal – gives minimum average waiting time for a given set of processes.

Shortest-Job-First (SJF) Scheduling

Process Arrival Time Burst Time P1 0.0 7 P2 2.0 4 P3 4.0 1 P4 5.0 4

SJF (non-preemptive)

Average waiting time = (0 + 6 + 3 + 7)/4 = 4

Example of Non-Preemptive SJF

P1 P3 P2

73 160

P4

8 12

Process Arrival Time Burst Time P1 0.0 7 P2 2.0 4 P3 4.0 1 P4 5.0 4

SJF (preemptive)

Average waiting time = (9 + 1 + 0 +2)/4 = 3

Example of Preemptive SJF

P1 P3P2

42 110

P4

5 7

P2 P1

16

The exponential average formula can only estimate the length.

Can be done by using the length of previous CPU bursts, using exponential averaging.

Predicting Next CPU Burst

:Define 4.

10 , 3.

burst CPU next the for value predicted 2.

burst CPU of lenght actual 1.

1n

thn nt

.1 1 nnn t

actual length

Next CPU Burst

=0◦ n+1 = n

◦ Recent history does not count.

=1◦ n+1 = tn

◦ Only the actual last CPU burst counts.

Exponential Averaging

If we expand the formula, we get:n+1 = tn+(1 - ) n

n+1 = tn+(1 - )[ tn-1 +(1 - ) n-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.

Exponential Averaging

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

Priority Scheduling

SJF is a priority scheduling where priority is the predicted next CPU burst time.

Problem Starvation – low priority processes may never execute.

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

Priority Scheduling

Each process gets a small unit of CPU time (time quantum), 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.

Performance◦ q large FIFO◦ q small q must be large with respect to context switch,

otherwise overhead is too high.

Round Robin (RR)

Process Burst TimeP1 53 P2 17 P3 68 P4 24

The Gantt chart is:

Typically, higher average turnaround than SJF, but better response.

RR with Time Quantum = 20

P1 P2 P3 P4 P1 P3 P4 P1 P3 P3

0 20 37 57 77 97 117 121 134 154 162

Time Quantum vs Context Switch Time

Turnaround Time Varies With The Time Quantum

Ready queue is partitioned into separate queues:foreground (interactive)background (batch)

Each queue has its own scheduling algorithm:

foreground – RRbackground – FCFS

Multilevel Queue

Scheduling must be done between the queues.

◦ Fixed priority scheduling; (i.e., serve all from foreground then from background). Possibility of starvation.

◦ Time slice – each queue gets a certain amount of CPU time which it can schedule amongst its processes; i.e., 80% to foreground in RR

◦ 20% to background in FCFS

Multilevel Queue

Multilevel Queue Scheduling

A process can move between the various queues; aging can be implemented this way.

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

Multilevel Feedback Queue

Three queues: ◦ Q0 – time quantum 8 milliseconds◦ Q1 – time quantum 16 milliseconds◦ Q2 – FCFS

Scheduling◦ A new job enters queue Q0 which is served FCFS.

When it gains CPU, job receives 8 milliseconds. If it does not finish in 8 milliseconds, job is moved to queue Q1.

◦ At Q1 job is again served FCFS and receives 16 additional milliseconds. If it still does not complete, it is preempted and moved to queue Q2.

Multilevel Feedback Queue

Multilevel Feedback Queues

Advantages:◦ Flexibility – can be used to implement many

different policies◦ Good balance of priority and fairness (and it’s

tweakeable)

Disadvantages:◦ Complexity to code◦ Run-time efficiency

Multilevel Feedback Queues

CPU scheduling gets more complex when multiple CPUs are available.

Assume homogeneous processors within a multiprocessor.◦ Note: Future processors may not fit this model!

Schedules can be made either:◦ Symmetrically – each processor schedules itself

from a shared queue◦ Asymmetrically – one processor only in kernel

mode

Multiple-Processor Scheduling

Processor Affinity◦ When a process was running on CPU a, and is

migrated to CPU b, there is a penalty for cache/TLB/memory flush/re-load.

Load Balancing:◦ Processes should be scheduled fairly evenly

across multiple processors◦ Avoids wasting CPU cycles when there is

something that could be running

These two are both important, but they tend to fight each other.

Affinity vs. Balance

To avoid flushing the caches, each process will “preference” a CPU (or set of CPUs)◦ “Soft” affinity – can be moved if needed◦ “Hard” affinity – locked in on the current CPU

Implementation varies – depends on:◦ System architecture and available resources◦ Load balancing needs

Processor Affinity

Solutions: Run with a single ready queue

◦ If a CPU becomes idle, it just pulls the next process

◦ Not used in practice (SMP common) Push migration

◦ An OS task examines the load every so often, and if needed, pushes some processes from heavily loaded CPUs to lightly loaded ones.

Pull migration◦ An idle CPU pulls a process from another CPUs

queue.

Load Balancing

Most multi-core processors today also implement hardware simultaneous multithreading◦ i.e. Intel i7 (Nehalem) “Hyperthreading”

CPU is represented to OS as multiple logical CPUs – one for each “hardware thread”◦ i.e. UltraSPARC T1 8 cores, each with 4 threads

32 logical processors!

Multi-CORE Scheduling

OS must schedule the processes into the “logical processors”.◦ Typical scheduling algorithms (?)

CPU decides which “hardware thread” to run on each core◦ Coarse-grained parallelism rotate threads when

a long-latency instruction happens, like a load which misses cache

◦ Fine-grained parallelism “Round Robin” the hardware threads, often every clock cycle. This is very rare in practice

Multi-CORE Scheduling

Deterministic modeling:

◦ takes a particular predetermined workload◦ defines the performance of each algorithm for

that workload.

What you did in 265

Algorithm Evaluation

Queueing models – Little’s law◦ N = λ W

N = average queue length λ = average arrival rate W = average waiting time in queue

◦ e.g., if N = 8, λ = 2 W = 4

Algorithm Evaluation

Simulations◦ Clock◦ Random generation of:

Processes Burst times Arrival & departures rates

Algorithm Evaluation

Based on probability distributions:◦ Uniform◦ Exponential◦ Poisson◦ Empirical

Implementation

Algorithm Evaluation

CPU Scheduler Evaluation by Simulation

UNIX schedulers

UNIX schedulers

Bands of Priorities

Priority Formula

Solaris Scheduling

Solaris Dispatch Table

Linux pre-2.6 O(n) SchedulerO(n) scheduling a task takes O(n) where n = # of tasks

One runqueue for all processors in a symmetric multiprocessor system

- Task can be scheduled on any CPU - Good for load balancing

- Bad for memory cache movements, e.g., task previously on CPU1 is - run on CPU2 move cache1 to cache2

Single runqueue CPUs had to contend with shared lock.No preemption allowed higher priority process may have to wait.

Linux 0(1) 2.6 Scheduler

- Each CPU has its own runqueue (priority = 1 to 140)- Tasks are scheduled RR multilevel feedback paradigm.- Tasks whose TQ expires are moved to Expired runqueue

(priorities are recalculated)

SMP Tasks Indexed on Priorities

Linux 0(1) 2.6 SchedulerWhy O(1)?- Bitmap of priorities is read (each priority level points to process)- Since size of bitmap is 140, selection of process does not depend on the number of processes in the runqueue.

Active runqueue pointer- When active runqueue is empty, pointer is set to expired runqueue, i.e., expired runqueue now becomes active runqueue and vice versa- Each CPU (in an SMP system) sets locks on its own runqueues all CPUs can schedule without contention from other CPUs.

Linux 0(1) 2.6 Scheduler

Dynamic Priority Assignment- CPU bound processes are penalized (increased by 5 levels)- I/O bound rewarded (prioirity# decreased by 5 levels) I/O bound use CPU to set up I/O and is suspended give other processes a chance to execute altruistic

Heuristic for I/O or CPU bound category??- interactivity heuristic based on: time task executes compared with time it is suspended.- I/O bounds sleep time is large increase in interactivity metric rewarded.- Priority adjustments only applied to user processes.

Priority-based, pre-emptive. Dispatcher handles scheduling. Thread currently running is interrupted

when:◦(a) it terminates, or◦(b) higher priority thread arrives, or◦(c) time-quantum expires, or◦(d) does I/O

gives real-time threads priority

Windows XP Scheduling

Dispatcher uses multi-level priority scheme

Priorities are divided into two classes:◦ (a) variable (1-15)◦ (b) real-time (16-31)

thread running at priority 0 is used for memory mgmt.

Processes are rewarded/penalized for sleeping/using up CPU

Windows XP Scheduling

Windows XP SchedulingPriority Class

Relative Priority