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

VRSQRT28PD

Computes the approximate reciprocal square root of packed double-precision floating-point values. It calculates $1/\sqrt{x}$ for each double-precision element in the source operand with a relative error of less than or equal to $2^{-28}$.

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

DO NOT support LOCK

This instruction is available only in 64-bit mode or compatibility mode. It requires the AVX support feature to be enabled.

The result is an approximation; therefore, the precision of the result is limited. Users SHOULD follow this instruction with Newton-Raphson iterations if higher precision is required. If the input is $\pm\infty$, the result is $+0$. If the input is $\text{NaN}$, the result is the same $\text{NaN}$. If the input is $\pm 0$, the result is $\infty$.

Floating-point exceptions are handled as follows:

• #P: May be set.
• #Z: Not set.
• #D: Not set.
• #O: Not set.
• #U: Not set.
• #I: Set if the input is negative.