This content has not been validated; it was AI-generated following the x86-64 ISA Documentation generation process to maintain high quality. However, as no human has verified this data, errors may exist. If the accuracy of this data is critical, please consult the Intel SDM. The model used to generate this is gemma4:31b

MOVNTDQ

Moves 128 bits of data from the source to the destination using a non-temporal store, bypassing the cache hierarchy to minimize cache pollution.

The following table covers what the source and destinations can be:

DO NOT support LOCK

The destination memory address MUST be aligned on a 16-byte boundary. If the address is not aligned, the instruction will trigger a general protection exception (#GP). This instruction is available in 64-bit mode and compatibility mode.

To ensure data consistency and visibility to other processors or devices, a SFENCE instruction SHOULD be executed after MOVNTDQ stores. Because this instruction uses write-combining, stores are not guaranteed to be committed to memory in program order. Failure to use a memory fence MAY lead to data corruption or stale reads in multi-threaded environments.