The OpenD Programming Language

_mm256_rsqrt_ps

Compute the approximate reciprocal square root of packed single-precision (32-bit) floating-point elements in a. The maximum relative error for this approximation is less than 1.5*2^-12.

Meta