Thirty-one general-purpose registers versus x86's sixteen is a fundamental design difference: compilers spill far less on AArch64, part of why Apple Silicon and Graviton punch above their clock speeds. The dedicated lr register also changes how calls, leaf functions, and stack traces work compared to x86.