I am considering the Ryzen 3900X vs Intel i9 9900K.
For reference to some of the replies, the Ryzen 3900x & i9-9900k are ~$500 CPUs. This is clearly a high end build. OP, since you're pairing with a graphics card anyways you might be interested in the i9-9900KF, which is a tad cheaper since it's the same chip with the on board graphics disabled.
Remark #1:
That said, for almost all applications, single core performance is going to be the bottleneck. Most software isn't designed to be multi-threaded, and even things that are still fall victim to
Amdahl's law. Parallel computing is no panacea for computer performance: imagine the total number of tasks it takes to process a turn of civ. Some of these tasks can be computed at the same time, some can't
because they require information from previous tasks. Even if
50% of the workload was parallelized with no overhead cost and
infinite cores, you'd still only speed up performance by 2x.
TLDR even applications that support multithreading still have huge blocks of code that has to be run single core, and most applications that do support multithreading aren't leveraging unlimited numbers of cores. Higher clock speed (GHz) and how efficient the processor is at handling tasks (some optimization can be done so fewer clock cycles are needed for the same instructions) directly affect both single core performance and parallel workloads. You need a
lot of extra cores to be
utilized to start making a dent in being less efficient per core. While AMD certainly made leaps, intel
generally has shown they still have the most efficient core designs on the market.
Remark #2:
Everything I just said about how good those cores are at crunching numbers goes out the window when the CPU itself needs to go get more data from RAM, or, god forbid, from disk. In CPU terms doing this takes an eternity: A well greased stick of DDR4 2666 ram
might have a latency of 10 nanoseconds. Do you know how long that is? That i9-9900k can get around 5Ghz on a single core - a clock cycle of 0.2 nanoseconds. Literally getting a single byte of information from ultra high performance ram is going to cost you 500 CPU clock cycles. The way that CPUs get around this is by implementing what are called "caches," where it keeps some data nearby so it doesn't always have to go to the ram. The downside to a bigger cache is that it takes longer to access (this is true for any type of addressed memory) so what chipmakers did was have the CPU's primary on chip data pull from a cache, and then that cache of data pulls from a cache. Hence we see L2 and L3 cache numbers quoted on the spec sheet for CPUs. Like all things, bigger is better (up to a point.) This is the primary advantage of the Ryzen- its caches are huge- and likely the source of why it sometimes beats intel and sometimes loses, imo. (That and it's on a 7nm process instead of 14nm like intel's currently is. That smaller size greatly improves any design.)
A Side comment on future proofing
My initial thought was Ryzen because of higher core count and the fact that it is somewhat future proof when I upgrade my GPU with the X570 boards utilizing PCIE Gen 4. But intel always seems to get better gaming performance.
Future proofing is a funny subject. In my experience, going to the higher end of the range can often be a decent investment- such as, say, splurging to move to an i9 k series. But paying the further premium to move from i9-9900k to i9-9900X is often not worth it, because at the very top end you pay a huge premium to the chipmaker for marginal gain. (Edit: I mean by this that the absolute best chips are expensive because they are hard to make, not because they are so superior in performance.)
The real bet right now for AMD vs Intel at this price point really depends on whether you think intel's response to AMD is going to be a repeat of the 2000s when the CEO's stated goal was to "bury AMD." This 9th gen was pretty rushed out the door because ryzen has been a real wake up call to stop being lazy, but intel certainly has the resources to become untouchable again. For example: the ryzen is on a 7nm process where intel's is a 14nm. I would guarantee intel could produce a 7nm chip before AMD moves to a smaller process node, and just making the i9 in 7nm will give it a nice bump in performance. The reason this matters is because the two processors take different sockets; but if you'll buy a new mobo in a few years anyways then, it doesn't matter.
