Multiple Processor Systems

Multiple Processor Systems
Chapter 8
Tanenbaum & Bo, Modern Operating Systems:4th ed., (c) 2013 Prentice-Hall, Inc. All rights reserved.
Multiple Processor Systems
Figure 8-1. (a) A shared-memory multiprocessor. (b) A message-

passing multicomputer. (c) A wide area distributed system
Multiprocessor Hardware
Figure 8-2. Three bus-based multiprocessors. (a) Without caching.

(b) With caching. (c) With caching and private memories.
UMA Multiprocessors Using
Crossbar Switches
Figure 8-3. (a) An 8 × 8 crossbar switch. (b) An open

crosspoint. (c) A closed crosspoint.
Multistage Switching Networks (1)
Figure 8-4. (a) A 2 × 2 switch with two input lines, A and B,

and two output lines, X and Y. (b) A message format.
Multistage Switching Networks (2)
Figure 8-5. An omega switching network.
NUMA Multiprocessors (1)
Characteristics of NUMA machines which
distinguish them from other multiprocessors:
1.There is a single address space visible to all
CPUs.
2.Access to remote memory is via LOAD and
STORE instructions.
3.Access to remote memory is slower than
access to local memory.
Figure 8-6. (a) A 256-node directory-based multiprocessor
Figure 8-6. (b) Division of a 32-bit memory address into

fields. (c) The directory at node 36.
Multiprocessor Operating System Types
Each CPU Has Its Own Operating System
Figure 8-7. Partitioning multiprocessor memory among four CPUs,

but sharing a single copy of the operating system code. The boxes
marked Data are the operating system’s private data for each CPU.
Master-Slave Multiprocessors
Figure 8-8. A master-slave multiprocessor model.
Symmetric Multiprocessors
Figure 8-9. The SMP multiprocessor model.
Multiprocessor Synchronization (1)
Figure 8-10. The TSL instruction can fail if the bus cannot be locked.
These four steps show a sequence of events where the failure is
demonstrated.
Multiprocessor Synchronization (2)
Figure 8-11. Use of multiple locks to avoid cache thrashing.
Time Sharing
Figure 8-12. Using a single data structure for scheduling a

multiprocessor.
Space Sharing
Figure 8-13. A set of 32 CPUs split into four partitions,

with two CPUs available.
Gang Scheduling (1)
Figure 8-14. Communication between two threads belonging

to thread A that are running out of phase.
Gang Scheduling (2)
Gang scheduling has three parts:
1.Groups of related threads are scheduled as a
unit, a gang.
2.All members of a gang run at once on different
timeshared CPUs.
3.All gang members start and end their time
slices together.
Gang Scheduling (3)
Figure 8-15. Gang scheduling.
Interconnection Technology (1)
Figure 8-16. Various interconnect topologies. (a) A single switch. (b)

A ring. (c) A grid. (d) A double torus. (e) A cube. (f) A 4D hypercube.
Interconnection Technology (2)
Figure 8-17. Store-and-forward packet switching.
Network Interfaces
Figure 8-18. Position of the network interface boards

in a multicomputer.
Blocking versus Nonblocking Calls (1)
Figure 8-19. (a) A blocking send call.

(b) A nonblocking send call.
Blocking versus Nonblocking Calls (2)
Choices on the sending side:
1.Blocking send (CPU idle during transmission).
2.Nonblocking send with copy (CPU time wasted
for the extra copy).
3.Nonblocking send with interrupt (makes
programming difficult).
4.Copy on write (extra copy probably needed
eventually).
Remote Procedure Call
Figure 8-20. Steps in making a remote procedure call.

The stubs are shaded gray.
Distributed Shared Memory (1)
Figure 8-21. Various layers where shared memory can be implemented. (a)
The hardware. (b) The operating system. (c) User-level software.
Figure 8-22. (a) Pages of the address space distributed

among four machines.
Figure 8-22. (b) Situation after CPU 1 references page 10 and

the page is moved there.
Figure 8-22. (c) Situation if page 10 is read only and

replication is used.
False Sharing
Figure 8-23. False sharing of a page containing

two unrelated variables.
Graph-Theoretic Deterministic
Algorithm
Figure 8-24. Two ways of allocating nine

processes to three nodes.
Sender-Initiated Distributed
Heuristic Algorithm
Figure 8-25. (a) An overloaded node looking for a lightly loaded node
to hand off processes to. (b) An empty node looking for work to do.
Distributed Systems (1)
Figure 8-26. Comparison of three kinds of multiple CPU systems.
Distributed Systems (2)
Figure 8-27. Positioning of middleware in a distributed system.
Network Hardware
Ethernet
Figure 8-28. (a) Classic Ethernet. (b) Switched Ethernet.
The Internet
Figure 8-29. A portion of the Internet.

Network Services
Figure 8-30. Six different types of network service.
Network Protocols
Figure 8-31. Accumulation of packet headers.
Document-Based Middleware (1)
Figure 8-32. The Web is a big directed graph of documents.
For the browser to get the page
http://www.minix3.org/getting-started/index.html
1.Browser asks DNS for IP address of www.minix3.org
2.DNS replies with 66.147.238.215
3.Browser makes a TCP connection to port 80 on
66.147.238.215
4.Browser sends a request asking for the file
getting-started/index.html
5. The www.minix3.org server sends the file

getting-started/index.html
6. Browser displays all the text in
getting-started/index.html.
7. Browser also fetches and displays all images on the
page
8. The TCP connection is released
File-System-Based Middleware
Transfer Model
Figure 8-33. (a) The upload/download model.

(b) The remote access model.
The Directory Hierarchy
Figure 8-34. (a) Two file servers. The squares are directories
and the circles are files. (b) A system in which all clients have
the same view of the file system. (c) A system in which
different clients may have different views of the file system.
Naming Transparency
Common approaches to file and directory
naming in a distributed system:
1.Machine + path naming, such as /machine/path
or machine:path.
2.Mounting remote file systems onto the local
file hierarchy.
3.A single name space that looks the same on all
machines.
Semantics of
File Sharing
Figure 8-35. (a) Sequential

consistency. (b) In a distributed
system with caching, reading a file
may return an obsolete value.
Object-Based Middleware
Figure 8-36. The main elements of a distributed system

based on CORBA. The CORBA parts are shown in gray.
Coordination-Based Middleware (1)
Figure 8-37. Three Linda tuples
Coordination-Based Middleware (2)
A match occurs if the following three conditions
are all met:
1.Template and tuple have the same number of
fields.
2.Types of corresponding fields are equal.
3.Each constant or variable in the template
matches its tuple field.
Publish/Subscribe
Figure 8-38. The publish/subscribe architecture.
End
Chapter 8

Multiple Processor Systems

Uploaded by

Copyright:

Available Formats

Multiple Processor Systems

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

Multiple Processor Systems

Uploaded by

Copyright:

Available Formats

Multiple Processor Systems

Figure 8-1. (a) A shared-memory multiprocessor. (b) A message-

Figure 8-2. Three bus-based multiprocessors. (a) Without caching.

Figure 8-3. (a) An 8 × 8 crossbar switch. (b) An open

Figure 8-4. (a) A 2 × 2 switch with two input lines, A and B,

Figure 8-5. An omega switching network.

Figure 8-6. (a) A 256-node directory-based multiprocessor

Figure 8-6. (b) Division of a 32-bit memory address into

Figure 8-7. Partitioning multiprocessor memory among four CPUs,

Figure 8-8. A master-slave multiprocessor model.

Figure 8-9. The SMP multiprocessor model.

Figure 8-11. Use of multiple locks to avoid cache thrashing.

Figure 8-12. Using a single data structure for scheduling a

Figure 8-13. A set of 32 CPUs split into four partitions,

Figure 8-14. Communication between two threads belonging

Figure 8-15. Gang scheduling.

Figure 8-16. Various interconnect topologies. (a) A single switch. (b)

Figure 8-17. Store-and-forward packet switching.

Figure 8-18. Position of the network interface boards

Figure 8-19. (a) A blocking send call.

Figure 8-20. Steps in making a remote procedure call.

Figure 8-22. (a) Pages of the address space distributed

Figure 8-22. (b) Situation after CPU 1 references page 10 and

Figure 8-22. (c) Situation if page 10 is read only and

Figure 8-23. False sharing of a page containing

Figure 8-24. Two ways of allocating nine

Figure 8-26. Comparison of three kinds of multiple CPU systems.

Figure 8-27. Positioning of middleware in a distributed system.

Figure 8-28. (a) Classic Ethernet. (b) Switched Ethernet.

Figure 8-29. A portion of the Internet.

Figure 8-30. Six different types of network service.

Figure 8-31. Accumulation of packet headers.

Figure 8-32. The Web is a big directed graph of documents.

5. The www.minix3.org server sends the file

Figure 8-33. (a) The upload/download model.

Figure 8-35. (a) Sequential

Figure 8-36. The main elements of a distributed system

Figure 8-37. Three Linda tuples

Figure 8-38. The publish/subscribe architecture.

You might also like