Actually, if you read the results carefully, they are what is funny. How Intel can polish an "excrement" and make it look like a gem.
It is not a joke, though. Think of it this way. Let's say you were on Skylake-SP not so long ago. You have one or more server rooms packed with machines. Oops! Spectre/Meltdown etc. happens. Now you are in a pickle since your next upgrade cycle might not have been due until 2020 or later. You can start looking into validation of existing EPYC hardware, wait even longer to begin validation for Rome (hopefully AMD will send you some samples to begin validation pre-market availability), or you can get the architecturally-identical Cascade Lake-SP from the same vendor that sold you Skylake-SP with much shorter validation time. With Cooper Lake on the horizon, there are some shops that are going to choose it as well for the same reason. At least this time around, procurement officers who are NOT being goaded into an unplanned upgrade thanks to <insertsecuritythreatoftheweekhere> will be able to look at something like Rome, compare TCO with Cooper Lake, and maybe make the switch. Even if you were relatively happy with the performance of Skylake-SP, you have to admit that you can move to something like Rome and enjoy a much lower TCO (versus Cooper Lake). But Intel definitely fed on inertia and unplanned upgrade cycles. They will feed again, and it will keep happening until enough shops finally get tired of the expense involved running 14nm Intel CPUs.