[__arm_]vmaxnmaq_m[_f16]

Find the floating-point maximum number of the elements in the source operands, and store the result in the corresponding destination elements. It handles NaNs in consistence with the IEEE754-2008 specification, and returns the numerical operand when one operand is numerical and the other is a quiet NaN. The absolute variant takes the absolute values of the elements from the destination vector and compares them to the absolute values of the corresponding elements in the source vector. The larger values are stored back into the destination vector.