When running make with multiple jobs (-j#), we get a massive speedup as we can use more CPU cores. The cost is unfortunately that the output will become jumbled as commands are intermixed. This makes it difficult to follow what happens when we need to debug something. Fortunately newer make has support for buffering data and outputting it in a more readable order (likely to compete with ninja). It is documented here: https://www.gnu.org/software/make/manual/html_node/Parallel-Output.html It is unclear what version added this. Our 3.82 doesn't have it, but Fedora's 4.3 does.
