Benchmarks rarely tell the whole story. They're a good way to compare two chips that are nearly identical but for one difference (it isolates the effect of that one difference), but they lose utility when comparing two chips that a lot of differences. There are just too many differences between x86/64 and ARM that comparing benchmarks is pretty much useless.
One thing is benchmarks don't really tell the whole story for software/hardware optimization. For example, Intel has virtualization optimization built-in their chips, something called Intel VT-x (I think AMD has an equivalent called AMD-v or something). If software is written to take advantage of it, it makes virtualizing other OSs though VM or Docker way more efficient and faster - nearly native speeds are possible. As far as I know, ARM doesn't have anything like that, and that huge advantage isn't reflected in any speed score benchmark.