[omikun]

X86 compete based on clock speed for the longest time so they use cell libraries designed for higher that. This means the transistors are larger and less dense. Arm cores are targeted at energy efficiency first so they use denser cells that doesn’t clock as fast. The trade off is they can have larger reorder buffers and larger scheduling windows to squeeze better ipc. As frequency scaling slows but not so much density scaling you get better results going the slower route.