While interesting to read, the title should be "On the Costs of Linux Syscalls"
"I ran the benchmark on a heterogeneous set of hosts, i.e. different kernels, operating systems and configurations"
In the fine print: the "different OS:es" were two Linux Distributions which are (if I understand correctly) not even that far apart tech wise? (RHEL and Fedora).
I would like to see this extended to other processor families.