"This is simply not true even Skondris his own renderings do prove that the claim isn't true at all. Only one one of the scenes shown is really faster, the others don't even come close to the 30%."

The "over 30%" improvement claim is given as an average of results. It is true that three out of the four renders SKondris used as examples didn't come close to 30% faster (one was slower), but the other one was stunningly faster than 30%.

Indeed, when you add up the times SKondris provides in his examples in your link, you find that the total Bryce5.01 time for the four scenes was 6280 seconds, whereas the Bryce5.5 times added up to 2753 seconds. I'm not going to bother working out what percentage that is, but you should be able to tell instantly that it is way over twice as fast - on average.

SKondris's examples in your link demonstrate an improvement better than the claim...