Intel's are generally far more smooth in many games. The key issue is not average fps, but higher *
minimum* fps as a result of lesser-well threaded games (or even lesser-well threaded areas / maps that are within a generally heavily threaded game) that soak up IPC & MHz and end up sporadically bottlenecking on even 5GHz OC'd AMD's. Eg:-
Skyrim:-
i5-3570K @ 3.4GHz =
88 min / 145 avg
FX-8350 @ 4.0GHz =
39 min /
90 avg
FX-8350 @ 4.8GHz =
45 min /
107 avg
http://www.bit-tech.net/hardware/2013/06/12/intel-core-i5-4670k-haswell-cpu-review/5
http://media.bestofmicro.com/7/G/315196/original/CPU-Core.png
The issue is not "average fps are all above 60fps so it's OK", it's
39fps vs
88fps min fps that makes the i5 feel & play much more smoothly especially in "busy" city areas. With VSync enabled on a 60Hz monitor, i5 & FX-8350 will both show "60fps average" in most areas. But min fps will still be
60 i5 vs
39 FX-8350, which is very noticeable when gaming in terms of "smoothness".
Likewise in BF4, an FX-8320 is
49min and an i5-4570
60min. Both cost the same $160 at Microcenter. OC the former to 5Ghz to match, and you've now got almost
200w higher power consumption adding roughly $10-20 annual running costs (plus water cooling costs) at which point same fair TCO over 2-3 years is now more like FX-8320 @ 5GHz vs i5-4670K @ 4.5GHz, and the gap promptly widens again - and that's in a game that should theoretically be "smoother" for AMD's with +4 threads in use...
AMD's aren't totally horrible chips, but an i5 over an FX-8320 is very definitely worth it for up to 125% higher min fps and generally smoother gameplay in 99.9% of the 50,000 or so PC games written over the past 30 years. Few games load 4-8 cores equally (it's often a load of 70-95% / 60-90% / 20-50% / 15-45%, etc), and even those that can split the load across 8 cores are often not consistently well threaded even from one map to another within the same game. When one "main" FX core goes past 100% load with code that's not easily threadable, your min fps will dip whether you have 7 or 700 spare cores. Of course, the same is true for Intel when an i5 core goes past 100%, but having 60-65% faster IPC cores per clock, it happens far less often.
As much as some AMD people may not like to admit it, single-threaded benchmarks can actually give a very valid reflection of higher min fps in many games (and how smooth the game is when "spiking"), regardless of what the avg fps may be. And IPC is still king even on "next gen" games. With poorly threaded games, min fps are up to 50%, 60%, even 70% higher on stronger Intel cores even with similar avg fps. Conversely even with heavily threaded games, with 4 strong cores,
even massive under-clocking can have little negative effect on performance.
Just as my old AMD X2 felt more smoother than Intel's P4, today Intel's core architecture just feels much more smoother than even OC'd AMD's, and when you ignore avg fps and look objectively at the min fps "spikes" in many games it's obvious why.