Right ok, I ran the following benchmarks twice just to see a little more in real terms what clocking of these cards gives, took the average and here is what we got:
Default 5700XT Clocks (1755mhz-gpu/1750mhz-mem, boost 1905mhz)
Superposition Benchmark 1080p Extreme: Score:4,736, Min:28.41fps, Avg:35.43fps, Max:42.20fps
3DMark TimeSpy Default: Score:7,039, Graphics Score:8,383, Test1:57.17fps, Test2:46.01fps
3DMark Fire Strike Extreme: Score:10,300, Graphics Score:12,068, Test1:60.36fps, Test2:46.41fps
Overclcoked 5700XT Clocks (2079mhz-gpu/1900mhz-mem, boost 2079mhz, 1176mv-voltage)
Superposition Benchmark 1080p Extreme: Score:5,246 Min:31.39fps, Avg:39.24fps, Max:46.72fps
3DMark TimeSpy Default: Score:7,647, Graphics Score:9,467, Test1:65.39fps, Test2:51.71fps
3DMark Fire Strike Extreme: Score:, Graphics Score:13,502, Test1:67.75fps, Test2:51.79fps
So, based on that lot we get the following % increase in fps based on those overclocks:
Superposition Benchmark: Avg fps = 9.7% Increase
3DMark TimeSpy Default: Test1 fps = 12.6% Increase, Test2 fps = 11.0% Increase
3DMark Fire Strike Extreme: Test1 fps = 10.9% Increase, Test2 fps = 10.4% Increase
So, basically, looking at this, going from stock to overclocked can give you circa 10-12% better fps. I assume it'll go for games as well. However, that sounds great, but lets be honest, let put that into real figures. OK, you're running 60fps at default, that means around 66-67.2fps at overclocked speeds... is that worth the hassle? I don't know, suppose it's what we all do to try and push our gear to get the most out of it, but all the hassle, more power, clocking, stability testing and everything all for that? hahahaha...