Load and Store

Move a value between a register and memory.

Memory System

Semantics

rd = *(base + offset) and the reverse. The interesting difference is not the operation but the addressing modes: how much address arithmetic the instruction can absorb.

Architecture Instructions Expressed as How this architecture does it
x86 one instruction One mnemonic for both directions; which one it is depends on which operand is memory. The addressing mode is unusually rich, base + index*scale + displacement in a single instruction, so array indexing rarely needs separate arithmetic.
ARM one instruction Separate mnemonics, with register or immediate offsets and pre- and post-indexed forms that write the computed address back to the base register. A pointer walk is therefore one instruction per step rather than two.
RISC-V one instruction Access width is in the mnemonic and the only addressing mode is base plus a 12-bit signed offset. No index register, no scaling, no writeback: address arithmetic is deliberately a separate instruction, which is the clearest expression of the ISA's minimalism.
PowerISA one instruction Width and extension are both in the mnemonic, so LWZ is load word and zero-extend. Indexed forms such as LWZX take two registers, and update forms such as LWZU write the address back, giving PowerISA both of the modes RISC-V leaves out.