- What is Concurrency?
- The Benefits of Parallel Programming
- The Benefits of Distributed Programming
- The Minimal Effort Required
- The Basic Layers of Software Concurrency
- No Keyword Support for Parallelism in C++
- Programming Environments for Parallel and Distributed Programming
- Summary·Toward Concurrency
This chapter is from the book
This chapter is from the book
This chapter is from the book
1.5 The Basic Layers of Software Concurrency
In this book we are concerned with concurrency within the application as opposed to concurrency at the operating system level, or concurrency within hardware. Although the concurrency within hardware and the concurrency at the operating system level support application concurrency, our focus is on the application. For our purposes, concurrency occurs at:
-
Instruction level
-
Routine (function/procedure) level
-
Object level
-
Application level
1.5.1 Concurrency at the Instruction Level
Concurrency at the instruction level occurs when multiple parts of a single instruction can be executed simultaneously. Figure 1-3 shows how a single instruction can be decomposed for simultaneous execution.
)
Figure 1-3. Decomposition of a single instruction.
In Figure 1-3, the component (A + B) can be executed at the same time as (C – D). This is an example of concurrency at the instruction level. This kind of parallelism is normally supported by compiler directives and is not under the direct control of a C++ programmer.
1.5.2 Concurrency at the Routine Level
The WBS structure of a program may be along function lines, that is, the total work involved in a software solution is divided between a number of functions. If these functions are assigned to threads, then each function can execute on a different processor and if enough processors are available, each function can execute simultaneously. We discuss threads in more detail in Chapter 4.
1.5.3 Concurrency at the Object Level
The WBS of a software solution may be distributed between objects. Each object can be assigned to a different thread, or process. Using the CORBA (Common Object Request Broker Architecture) standard, each object may be assigned to a different computer on the network or different computer on a different network. We discuss CORBA in more detail in Chapter 8. Objects residing in different threads or processes may execute their methods concurrently.
1.5.4 Concurrency of Applications
Two or more applications can cooperatively work together to solve some problem. Although the application may have originally been designed separately and for different purposes, the principles of code reuse often allow applications to cooperate. In these circumstances two separate applications work together as a single distributed application. For example, the Clipboard was not designed to work with any one application but can be used by a variety of applications on the desktop. Some uses of the Clipboard had not been dreamed of during its original design.
The second and the third layers are the primary layers of concurrency that we will focus on in this book. We show techniques for implementing concurrency in these layers. Operating system and hardware issues are presented only where they are necessary in the context of application design. Once we have an appropriate WBS for a parallel programming or distributed programming design, the question is how do we implement it in C++.