In this talk we will discuss detailed analysis of the Yocto build on the resource-per-tool level. Understanding where the time goes in a build and what are the bottlenecks in your implementation can help with choosing the right sizing for your build machine or build grid.
Using the example of an Automotive Grade Linux build, We will analyze actual CPU time spent on compiling, packaging and performing other common tasks, gaining insights on what makes builds take as much as they do, which build-time resource constraints exist, and how they could be alleviated.

In this talk we will discuss how sstate-cache works and in which scenarios it shines. We will look at how it can coexist with another caching layer on the compiler level, and which in which scenarios each of the cache implementations can be useful to cut down the amount of computational work that needs to be done when doing incremental builds.