Round a single-precision floating-point number to the nearest power of 10 on a linear scale. - View it on GitHub