The last column is the most important as it shows the typical value, and these values are close. One of the hosts, probably, has something different in the settings -BIOS, OS tuning, some daemon running, management interrupts and it cause to spikes in some iterations.
Try to run loopback tests.
