Book Image

Software Architecture with C++

By : Adrian Ostrowski, Piotr Gaczkowski
Book Image

Software Architecture with C++

By: Adrian Ostrowski, Piotr Gaczkowski

Overview of this book

Software architecture refers to the high-level design of complex applications. It is evolving just like the languages we use, but there are architectural concepts and patterns that you can learn to write high-performance apps in a high-level language without sacrificing readability and maintainability. If you're working with modern C++, this practical guide will help you put your knowledge to work and design distributed, large-scale apps. You'll start by getting up to speed with architectural concepts, including established patterns and rising trends, then move on to understanding what software architecture actually is and start exploring its components. Next, you'll discover the design concepts involved in application architecture and the patterns in software development, before going on to learn how to build, package, integrate, and deploy your components. In the concluding chapters, you'll explore different architectural qualities, such as maintainability, reusability, testability, performance, scalability, and security. Finally, you will get an overview of distributed systems, such as service-oriented architecture, microservices, and cloud-native, and understand how to apply them in application development. By the end of this book, you'll be able to build distributed services using modern C++ and associated tools to deliver solutions as per your clients' requirements.
Table of Contents (24 chapters)
1
Section 1: Concepts and Components of Software Architecture
5
Section 2: The Design and Development of C++ Software
6
Architectural and System Design
10
Section 3: Architectural Quality Attributes
15
Section 4: Cloud-Native Design Principles
21
About Packt

Parallelizing computations using OpenMP and MPI

An alternative to using the standard parallel algorithms would be to leverage OpenMP's pragmas. They're an easy way to parallelize many types of computations by just adding a few lines of code. And if you want to distribute your code across a cluster, you might want to see what MPI can do for you. Those two can also be joined together.

With OpenMP, you can use various pragmas to easily parallelize code. For instance, you can write #pragma openmp parallel for before a for loop to get it executed using parallel threads. The library can do much more, such as executing computations on GPUs and other accelerators.

Integrating MPI into your project is harder than just adding an appropriate pragma. Here, you'll need to use the MPI API in your code base to send or receive data between processes (using calls such as MPI_Send and MPI_Recv), or perform various gather and reduce operations (calling MPI_Bcast and MPI_Reduce, among...