The poster explains the poor scaling. It is probably due to Amdahl's Law, a good portion of sequential code that is not speeded up by parallelisation.
"The speedup in the fixed section of the code is likely a fair bit higher than what's shown here."
But, yes, the case shows how easy it is to introduce a scaling problem if the programmer does not understand multi-threading and shared memory well. Luckily, the shared counter, in this case, was useless and could be removed.
PS. Pricey graphics cards from Nvidia may actually help AMD CPU sales in the gaming segment, as wise gamers know that they can achieve better performance by saving a little on CPU and put the savings towards a high-end GPU. From the same thread at Overclockers forums:
"I just don't know what to do. For a purely gaming rig I know that Intel mainstream will likely produce the best performance. I remember switching from a 6700k to a 5820k thinking I'd notice an improvement and I didn't, like at all. I imagine the 7820x would be that scenario all over again, no improvement at all with 2 extra cores over the mainstream i7. I'd be lying if I said I wasn't tempted by Ryzen, but I suffer from buyers remorse, and even if I was only 10fps short on a benchmark, I'd be wishing I had the cpu with better results (as silly as it sounds). However with my budget, I could build a Ryzen rig that would allow me to use the difference saved and pick up a 1080ti. The sensible part of me realises that a 1700 and 1080ti is going to be better than an i7 and 1080 at higher resolutions. Infact using my intel logic above I even thought about saving even more and going with a 1600, as 8 cores just aren't needed for gaming (by the time they are we'll be 3 or 4 cpu generations forward at least), and we know for certain the 8700k isn't going to be anywhere near the Ryzen 5 in price....Decisions, decisions.... "