Just a couple of points:
The G3258 is probably fine with major overclocks with just the stock cooler (i.e. the one it comes with). It's the same cooler as comes with Intel's top-tier CPUs, which have much higher TDPs than the G3258. It takes a pretty large OC on an unlucky chip to take the heat output of the G3258 above the 90W the cooler is rated for.
Bus width isn't directly proportional to performance - the whole point of nVidia's Maxwell architecture is to use things like busses intelligently so as to make do with narrow ones in order to reduce power consumption by reducing component counts. Don't dismiss a GPU because it's got a narrow memory bus. Instead, look at benchmarks - if a GPU beats another in a wide range of games, it's faster.
The G3258 is probably fine with major overclocks with just the stock cooler (i.e. the one it comes with). It's the same cooler as comes with Intel's top-tier CPUs, which have much higher TDPs than the G3258. It takes a pretty large OC on an unlucky chip to take the heat output of the G3258 above the 90W the cooler is rated for.
Bus width isn't directly proportional to performance - the whole point of nVidia's Maxwell architecture is to use things like busses intelligently so as to make do with narrow ones in order to reduce power consumption by reducing component counts. Don't dismiss a GPU because it's got a narrow memory bus. Instead, look at benchmarks - if a GPU beats another in a wide range of games, it's faster.
OS: Windows 10 64 bit Professional
CPU: AMD Ryzen 5900X
RAM: 16GB
GPU: Radeon Vega 56
CPU: AMD Ryzen 5900X
RAM: 16GB
GPU: Radeon Vega 56