Code-Tuning Techniques

[[deleted user]]

I don't recall ever seeing this careful of an analysis, done with multiple programming languages, assessing the impact of common code tweaking techniques. It vividly illustrates the futility of trying to guess the benefit (if there even is one) of a tweak in advance. As stated over and over again in this chapter, you have to measure it to find out if it is worthwhile. I was really surprised at many of the results that appear here.

This chapter also serves as a nice little guide to the kinds of optimizations that compilers routinely automate.

[Erik]

Yes, this chapter has a good theme about not guessing the benefits. I'm not a big fan of unrolling loops, but it certainly can make things faster.

Being an optimization expert in multiple languages seems very challenging. Good thing there are several types of profiling tools to help remove the guess work.

[Aleksander]

I find examples and data in this chapter as interesting as in previous chapter. It could be interesting to measure complexity and performance time and analyze the dependency of one on another. Some of the data seems reasonable, the other not so much, but it’s hard to argue with statistics...