I am trying to calculate mathematically here, but compared to ARM64 devices, X86 devices have more IPC. Thus breaks all the math. Can't compare ARM to x86 then. Does anyone here use it? How much speed can you safely have?
The N100 has almost the same performance as a Core i5-6500T. It can route 10G in software with no issues.
But what about SQM?
SQM on any remotely modern x86 box would barely affect performance. You'd only run into issues at 10G if using any packet dissector/IDS at line speeds.
Maybe have a look at the IPC of recent apple arm cores... before you generalize... sure mid- to high-end x86 CPUs tend to outcompete the ARM cores typically used in cheapish routers, but that is more a testament to the shitty CPUs vendors put into routers...
Also keep in mind, that the CPU alone is not enough you also need a sufficiently beefy memory and IO system to actually feed that CPU , and again a field where router makers tend to drop the ball a bit.
Mind you, the router makers and their SoC suppliers are not stupid, but they clearly bet on offload engines to do the heavy lifting relegating the CPUs to control plane duty (similar to what big iron routers do). But SQM typically is not one of the offered offloads...
Sorry, to get SQM with its traffic shaper to operate at 10 Gbps you need sufficiently beefy hardware, and I am not sure N100 and friends will do... for 1 Gbps and maybe 2.5 N100 should do, if configured competently. However, I neither have 1 Gbps* WAN nor N100 in personal use so I only offer second hand expertise here.
I have a passive cooled N100 router with 4 x 2.5G ports and use SQM with Cake. It is not able to handle 2.5Gbps with cake, 1Gbps yes. There may be little differences depending on the unit, as you can also get N100 units that are cooled with a fan. But for 2.5G shaping you better get something like this: