Book Image

Delphi High Performance

By : Primož Gabrijelčič
Book Image

Delphi High Performance

By: Primož Gabrijelčič

Overview of this book

Delphi is a cross-platform Integrated Development Environment (IDE) that supports rapid application development for Microsoft Windows, Apple Mac OS X, Google Android, iOS, and now Linux with RAD Studio 10.2. This book will be your guide to build efficient high performance applications with Delphi. The book begins by explaining how to find performance bottlenecks and apply the correct algorithm to fix them. It will teach you how to improve your algorithms before taking you through parallel programming. You’ll then explore various tools to build highly concurrent applications. After that, you’ll delve into improving the performance of your code and master cross-platform RTL improvements. Finally, we’ll go through memory management with Delphi and you’ll see how to leverage several external libraries to write better performing programs. By the end of the book, you’ll have the knowledge to create high performance applications with Delphi.
Table of Contents (16 chapters)
Title Page
Copyright and Credits
Packt Upsell
Contributors
Preface
Index

Pipelines


So far, I have discussed two different categories of problems that can be solved using patterns. In the first category, there are long operations that we just want to push into the background. We are happy with the execution time, but we would like to keep the user interface responsive, and so we need to execute them in a background thread. Async/Await and Future are tools for this occasion.

In the second category, there are problems that can be split into parts that are independent or mostly independent of each other. We can use Join, Join/Await, or Parallel for to parallelize them and implement the "mostly" part with locking or interlocked operations.

This, however, doesn't even nearly cover all the use cases. There's at least one big category left that is hard to parallelize with the Parallel Programming Library tools. I'm talking about problems that are hard to split into parts but where operations executed on a given data item (an atomic part of the input data) can be split...