First, let's have a look at what happens when we run only the games with no other network resource-stealing applications in the background:
Surprisingly, we got nearly-identical latencies and frame rates. The Killer Xeno Pro's method of bypassing the Windows network stack doesn't seem to effectively lower latencies when there are no other programs fighting for network bandwidth.
Let's see how both options affect frame rates:
Notice the fluctuations in frame rates. They are close, but even the Killer Xeno Pro loses a few in the TF2 benchmark and gains a couple in the WoW benchmark. Remember, these types of situations are less than ideal for frame rate testing because no two tests will deliver identical visual output. As a trend, we can see that there's probably no real-world difference as the variances that we see are well within a good margin of error.
As far as the subjective tests go, we noticed no difference between the Killer Xeno Pro and the integrated NIC as far as how it felt and how we perceived the smoothness of the game play.