Manual
Programming Reference
rg16_float
Ordinary format with two components stored as 16-bit (half-precision) floating-point numbers.