CPU Scheduling

1
CPU Scheduling
2
Scheduling Task vs I/O Request

• Task computation time unpredictable
• Task can be preemptive

3
CPU Scheduling Algorithm

• FCFS
• Priority-based
• Proportional fair-share
• EDF
• Rate monotonic

4
Rate Monotonic
5
EDF
6
Problem

• How can multimedia applications co-exists with
  normal applications?

7
Same Idea as Cello

• P. Goyal, X. Guo, and H. M. Vin.
• A hierarchical CPU scheduler for
• multimedia operating systems.
• OSDI96

8
Guarantee?

• So far Best-Effort Real-Time Scheduling
• What about Guaranteed service?

9
How to Guarantee Services?
10
Resource Reservation CPU
11
Resource Reservations

• How to reserve?
• Overbook?
• Overuse?

12
Example Memory

• How to reserve?
• malloc
• Overbook?
• return NULL
• Overuse?
• crash, throw exception

13
Example CPU

• How to reserve?
• Overbook?
• Overuse?

Memory is discrete. CPU time is continuous.
14
How to Reserve?
15
CPU Reservation

• I need C units of time ..
• out of every T units.

16
Some Issues

• I need C units of time, out of every T units.

17
Effects of T and C
18
Effects of T and C
19
Schedulability Theorems

• Rate Monotonic Scheduler
• Earliest Deadline First

20
Assumptions

• programs are periodic
• computation time is constant
• zero context switch time
• programs are independent

21
Rialto Scheduler

• CPU Reservations and Time Contraints Efficient
  and Predictable Scheduling of Independent
  Activities,
• M. Jones, D. Rosu and M. Rosu

22
Why Rialto?

• Provide CPU reservation
• Enable deadlines
• Co-exists with best-effort apps

23
Overview

• input a set of tasks and their reservation
• pre-compute a scheduling graph
• schedule tasks based on scheduling graph

24
E
B
C
A
E
B
F
D
E
B
A
E
B
D
25
C
A
E
B
F
D
26
Advantages

• Only need to make new scheduling decisions when
  new task is added

27
Computing Scheduling Graph

• Input A set of tasks with their reservations
• Output A scheduling graph if feasible

28
Many Possible Solutions
C
A
A
E
B
E
B
F
F
D
D
C
29
C
A
F
E
B
F
D
30
Computing Scheduling Graph

• Input A set of tasks with their reservations
• Output A scheduling graph if feasible
• Goals (a) Minimize context switches (b)
  Distribute free time evenly

31
Life is Tough..
NP-Hard
32
Heuristic
10
10
10
10
10
10
10
33
10
5
F
5
5
10
10
10
34
6
1
F
1
BE
1
6
6
6
35
C
4
1
F
1
BE
1
1
1
AD
1
36
Time Constraint

• I need to run the following code, that takes
  30ms, 100ms from now, and the deadline is
  (now180ms).

37
Pseudocode

• can begin_constraint(start_time, deadline,
  estimate)
• if can schedule then
• do work
• else
• adapt
• time_taken end_constraint()

38
Scheduling Constraints
C
A
F
E
B
F
D
39
Advantages

• Only need to make new scheduling decisions when
  new task is added
• Feasibility of time constraint can be predicted
  accurately

40
Running Unreserved Task
C
A
F
E
B
F
D
41
Media Player on Rialto/NT
42
Look inside NT Scheduler

• Scheduling unit Thread
• Priority level 1-15, 16-31
• rt 24 high 13 normal 8 idle 4

43
Priority Boosting

• Example
• Priority 6 after awaken by keyboard input

44
Anti-Starvation

• If hasnt run for 3 seconds
• Boost priority to 14
• Avoid Priority Inversion

45
Mars Pathfinder
46
Rialto/NT
Rialto
Boost Priority to 30
NT Scheduler
47
Win Media Player 6.4
48
Experiment 1

• WMP running alone

49
Experiment 2

• WMP running with priority 8 competitor

50
Experiment 3

• WMP running with priority 10 competitor

51
Experiment 5

• WMP running with priority 10 competitor

52
Experiment 6

• WMP running with priority 10 competitor

53
Experiment 7

• WMP running with priority 10 competitor

54
Summary
55
CPU Scheduling

• With guarantee
• percentage of CPU time
• task with deadline
• Algorithms
• Rate-monotonic
• EDF
• Rialto