It's not just a CPU usage issue, either— vmmem leaks and holds onto several times as much memory as the actual guests are using, inevitably chewing up whatever you've set as your maximum quota for WSL even when the guests are only using a few hundred megs of RAM.