It's not going to happen, and why is a fundamental aspect of why AMD64 (x86) is different to ARMv7/v8/v9.

ARM gets parallelism at the instruction level (ILP), it has small instructions, very suited for out of order execution. A single thread can get a lot of parallelism, so the CPU doesn't need a lot of overhead to wrestle the ILP bear. The ease at which ARM (and some other RISC archs, like POWER) gets ILP is part of the reason why mandated parallelism in things like VLIW and EPIC didn't do very well: It just wasn't necessary.

AMD64 is very different, it has all that x86 baggage on it. Instructions have all kinds of modes, dependencies, tags, etc., and this makes them a lot more interdependent than ARM instructions are. So, pulling ILP out of AMD64 is a lot more difficult than ARM, and the CPU has to spend a lot more resource in doing it. Even then, it doesn't get the same degree of ILP ARM can achieve.

AMD64 gets more of its parallelism from TLP: Thread level parallelism. There's a reason all performance AMD64 processors from AMD and Intel support simultaneous multithreading (SMT/HyperThreading^(tm)), this is where most of their parallelism comes from. SMT shows a significant performance improvement in almost all cases, meaning execution slots are going spare when SMT isn't in use, which further means there isn't enough ILP to saturate the processor's capability.

This isn't usually the case on ARM, most ARM cores are designed to "race to sleep" and fill as much execution resource as possible. The CPU's awake and clocked up, it darned well better use that time as productively as possible, there's a power budget to worry about!

So, while-ever we're using AMD64, which will be more or less forever as far as the immediate future is concerned, similar performance on AMD64 will always need more power than it will on ARM.