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

VPERM2I128

This instruction permutes 32-bit elements from two 128-bit source operands based on an index provided in a third operand. The elements are selected from the combined 256-bit pool of the two sources and placed into the destination register.

The table after the description covers what the source and destinations can be.

DO NOT support LOCK

This instruction SHALL only be executed when the AVX-512 VBMI (Vector Byte Manipulation Instructions) subset is supported by the processor. It is NOT supported in compatibility mode if the CPU requires 64-bit mode for AVX-512 extensions.

To avoid undefined behavior, the index provided in the control register MUST be within the range of 0 to 7 for each 32-bit lane of the destination. Indices outside this range SHALL result in the corresponding element in the destination being set to zero.