Chapter 10. Improving the Performance of make

make plays a critical role in the development process. It combines the elements of a project to create an application while allowing the developer to avoid the subtle errors caused by accidentally omitting steps of the build. However, if developers avoid using make, because they feel the makefile is too slow, all the benefits of make are lost. It is important, therefore, to ensure that the makefile be crafted to be as efficient as possible.

Performance issues are always tricky, but become even more so when the perception of users and different paths through the code are considered. Not every target of a makefile is worth optimizing. Even radical optimizations might not be worth the effort depending on your environment. For instance, reducing the time of an operation from 90 minutes to 45 minutes may be immaterial since even the faster time is a "go get lunch" operation. On the other hand, reducing a task from 2 minutes to 1 might be received with cheers if developers are twiddling their thumbs during that time.

When writing a makefile for efficient execution, it is important to know the costs of various operations and to know what operations are being performed. In the following sections, we will perform some simple benchmarking to quantify these general comments and present techniques to help identify bottlenecks.

A complementary approach to improving performance is to take advantage of parallelism and local network topology. By running more than one command script at a time (even on a uniprocessor), build times can be reduced.