Why so slow?

This project has been optimized to run on a very beefy (196 core / 64 gb) cloud desktop. In that environment it uses ~5 cpu hours to run. However on smaller machines (like my personal 32 core / 64 gb desktop) it runs incredibly slowly. @tareknaser reports that is even slower on his MacBook. We should investigate why. Is it running out of RAM? Is there thread contention? Something else?