Basically, the code handler ordering does not generalise well across benchmark nor processor, so to get the speedup they see, you'd need a specialised interpreter for your specific benchmark and processor. That puts this into the "interesting, but not very practical" category.
This case is handled correctly. SVE loops don't count up a fixed number of elements per loop iteration, instead they count up the number of elements set in the predicate register.
If you perform a page-crossing load where, say, only elements 0-5 out of 32 elements are loaded, then only elements 0-5 in the predicate register gets set, the loop steps 6 elements instead of 32 elements in this iteration. On the next time around, if you really intended to go past the end, byte 0 of the next page will be in the first element position and goes through normal fault handling.
Lemire's example loop is subtly wrong here, but the version in the SVE slides he links to is fine.
Others have discussed the slavery claims, I'll address the linguistic claims.
The three-letter root عبد, transliterated "3abd", or more inaccurately, "Abed", has meanings related to servant and slaves. It does not mean black. This can be seen in this entry of the Hans Wehr dictionary - http://tinyurl.com/pk2alcu . You might be familiar with this root from the Arabic male name Abdallah or 3abdallah - this literally means "Servant of God".
Side note - I'm using the letter 3 to indicate the http://en.wikipedia.org/wiki/Ayin letter, which is a throaty consonant in Arabic that is hard for Westerners to pronounce and hear. Because of this, it is sometimes sliced off and ignored when Arabic words are transliterated to the Latin alphabet. However, making claims about the Arabic language while ignoring it is a bit like making claims about the etymology of English words while ignoring the difference between l and r - it can lead you seriously astray.
The color black is أسود / aswad (masculine) or سوداء / sauda2 (feminine) in Arabic, all derived from the three-letter root سود (s-u-d). Dictionary entry at http://tinyurl.com/q7cadg4 . You may recognize this root from the country سودان / Sudan. However, it should not be confused with Saudi Arabia, which is سعودية or sa3udia. Again, the Ayin is present in the word and gives it a very different meaning.
In general, Arabic has a number of consonants that are hard to distinguish for Westerners - this includes the Ayin, the gh (guttural r), the difference between aspirated and non-aspirated h, the difference between emphatic and non-emphatic t, s, d and dh. This means it is hard to represent Arabic accurately with the Latin alphabet without extending it in ways unfamiliar to regular readers. I tend to trust claims about Arabic if the Arabic alphabet is used to explain them, otherwise I would be fairly skeptical.
If that's the patent in question, it's so specific as to be useless.
The first claim, which all other claims depend on, only claims the invention when
1) The voting patterns are being displayed in a "swarm interface"
2) The "thickness of a line" is used for visualizing the number of votes
I don't believe that any of the modern sites use a swarm interface at all, and if someone wanted to show a swarm interface, by using something other than the thickness of a line for the visualization, you've sidestepped the patent.
It's hard to believe someone paid good money for this.
Basically, the code handler ordering does not generalise well across benchmark nor processor, so to get the speedup they see, you'd need a specialised interpreter for your specific benchmark and processor. That puts this into the "interesting, but not very practical" category.