I hope 2017 will be the year where dual-AMD computers (as in both processor and graphics card are AMD) become common again. Gaming rigs are quite overpriced right now because there doesn't seem to be enough competition from AMD. If AMD Zen is as good as advertised, then maybe we can get very a decent machine to play Civ VI at the highest settings at reasonable prices.
I do hope Zen is as good as advertised. Without competition, Intel has been delivering bad value to consumers, IMHO. If the rumors are to be believed, the low-end Zen will be 4C/8HT. It remains to be seen how effective their HT solution is. Intel's solution is proprietary, and AMD in recent years has not demonstrated the ability to deliver equally efficient chip designs. But, if the Zen can compete, and AMD undercuts Intel pricing, that will translate to more and more consumers having 8+ logical cores. It's not unreasonable to think that 2017 will be the year that developers start targeting multi-core PCs in greater numbers.
Even in 2016 it's a big jump in the number of titles. Battlefield 1 is HT-friendly until you hit 8 cores, as is RotTR. So is Doom in Vulkan mode. (If you want the framerate locked at 144Hz and you aren't rocking a fast CPU)
Add in the number of titles shipping this holiday season with DX12 support, and the picture should look pretty different next year.
Ironically, this is bad news for chip makers. There are millions of PCs out there with idle cores, and games are shipping more CPU-efficient. With less reason than ever to upgrade your CPU, AMD may be re-entering a very bad market.
EDIT: I stand corrected on 8-core market penetration. The steam report only shows physical cores. So it's probably more accurate to say that only 25% of PC owners have 8 logical cores. And yes, the typical performance improvement (when the # of physical cores is insufficient for the title) is 20%-30%, 50% in certain circumstances. So, it is true that if you take any two otherwise identical chips, where one is running in 3C/0HT mode, and the other is running in 2C/4HT mode, the 3C will usually win. An otherwise identical 6C/0HT would generally outperform a 4C/8HT. If you can afford an 8-core, by all means run it with hyper-threading turned off, because it will likely be a while before most titles take advantage of 12 logical cores. (Which is about what it would take before a 6C/12HT outperformed an 8C/0HT)
However, Civ6 could very well be the biggest exception to the rule. With peak usage of 16-20 logical cores during AI calculations, you're probably better off with the high-core count hyper-threaded machine, *as long as you're content with the average FPS*.
For my 6-core, the framerate almost never drops below 60, so hyper-threading is worth it to get the improved "next turn" processing. My 12-core (which can't be overclocked) can't quite hold 60 with the current DX11 implementation, due to the thermal throttling and variable turbo. But it screams through the AI calculations. It seemed to perform best with 8C/16HT and 12C/0HT, but it was hard to measure since the benchmark is measuring both average FPS and AI turn times. I supect 12C/24HT would have scored better if half the benchmark wasn't bottlenecked by the render threads.
Anyway, I'll probably have to wait for DX12 to play on that machine.
DRAM considerations make it very complicated. Skylake, with only dual-channel DDR4, has literally half the memory bandwidth of Haswell-E. Once you max out that memory bandwidth, threads will stall. This is true whether you're hyper-threaded or not. So if there's another thread running on that core that can work with in-cache memory while that other thread is stalled, hyper-threading will perform better. Without hyper-threading, that stalled thread won't relinquish the CPU, so that core just does nothing for a while. But, generally speaking, more threads = more DRAM pressure. There is a point of diminishing returns. (The cure for memory stalls is not more memory reads...)
I'm quite excited about the possibilities of high-end Zen APUs with HBM2. That could be a game-changer.