Yeah, I saw that one I am more concerned with the fact people believe the 3600(non X) will outperform the i9-9900K
The massive amounts of L3 cache available per core (especially in the case of the 6-core variants) may make it possible for some benchmarks to run almost entirely from within cache, which dramatically improves performance. We've seen this happen with Geekbench as well, which is known to benefit a lot from high memory bandwidth. These are obviously edge cases but not necessarily untrue. It simply confirms once again that many synthetic benchmarks do not represent real world performance.
One thing I noticed was that nearly every benchmark of 3rd gen Ryzen I've seen so far had a suboptimal memory configuration, running at 2133-2400 MT/s, or 4266 MT/s, which runs in 2:1 mode. There are also some very obvious problems with write throughput, which used to be Zen's strong point and is supposed to be even better for Zen 2. Are we looking at early and unoptimized BIOS versions?