Working with loops

So far, we have learned about several IR units such as modules, functions, basic blocks, and instructions. We have also learned about some logical units such as CFG and call graphs. In this section, we are going to look at a more logical IR unit: a loop.

Loops are ubiquitous constructions that are heavily used by programmers. Not to mention that nearly every programming language contains this concept, too. A loop repeatedly executes a certain number of instructions multiple times, which, of course, saves programmers lots of effort from repeating that code by themselves. However, if the loop contains any inefficient code – for example, a time-consuming memory load that always delivers the same value – the performance slowdown will also be magnified by the number of iterations.

Therefore, it is the compiler's job to eliminate as many flaws as possible from a loop. In addition to removing suboptimal code from loops, since loops are on the critical...