handouts/src/Advanced/Fast Inverse Root/parts/7 quake.tex

\section{The Fast Inverse Square Root}

The following code is present in \textit{Quake III Arena} (1999):
\lstset{
	breaklines=false,
	numbersep=5pt,
	xrightmargin=0in
}
\begin{lstlisting}[language=C]
float Q_rsqrt( float number ) {
    long i = * ( long * ) &number;
    i = 0x5f3759df - ( i >> 1 );
    return * ( float * ) &i;
}
\end{lstlisting}

This code defines
a function \texttt{Q\_rsqrt} that consumes a float named
\texttt{number} and approximates its inverse square root (in other words, \texttt{Q\_rsqrt} computes $1/\sqrt{\texttt{number}}$).

\vspace{2mm}

If we rewrite this using the notation we're familiar with, we get the following:
\begin{equation*}
	\texttt{Q\_sqrt}(n_i) = 6240089 - (n_i \div 2) \approx \frac{1}{\sqrt{n_f}}
\end{equation*}
$6240089$ is the decimal value of hex \texttt{0x5f3759df}. \par
It is a magic number hard-coded into the function.

\vspace{2mm}

Our goal in this section is to understand why this works. \par
How are we able to approximate $\frac{1}{\sqrt{x}}$ by only subtracting and dividing by two??

\problem{}
Using basic log rules, rewrite $\log_2(1 / \sqrt{x})$ in terms of $\log_2(x)$.

\begin{solution}
	\begin{equation*}
		\log_2(1 / \sqrt{x}) = \frac{-1}{2}\log_2(x)
	\end{equation*}
\end{solution}


\vfill


\problem{}
Find the exact value of $\kappa$ in terms of $\varepsilon$. \par
\note{Remember, $\varepsilon$ is the correction term in the approximation $\log_2(1 + a) = a + \varepsilon$.}

\begin{solution}
	Say $g_f = \frac{1}{\sqrt{n_f}}$ (i.e, $g_f$ is the value we want to compute). We then have:
	\begin{align*}
		\log_2(g_f)
		&=\log_2(\frac{1}{\sqrt{n_f}}) \\
		&=\frac{-1}{2}\log_2(n_f) \\
		&=\frac{-1}{2}\left( \frac{n_i}{2^{23}} + \varepsilon - 127 \right)
	\end{align*}
	But we also know that
	\begin{align*}
		\log_2(g_f)
		&=\frac{g_i}{2^{23}} + \varepsilon - 127
	\end{align*}
	So,
	\begin{align*}
		\frac{g_i}{2^{23}} + \varepsilon - 127
		&=\frac{-1}{2}\left( \frac{n_i}{2^{23}} + \varepsilon - 127 \right) \\
		\frac{g_i}{2^{23}}
		&=\frac{-1}{2}\left( \frac{n_i}{2^{23}} \right) + \frac{3}{2}(\varepsilon - 127) \\
		g_i
		&=\frac{-1}{2}\left(n_i \right) + 2^{23}\frac{3}{2}(\varepsilon - 127) \\
		&=2^{23}\frac{3}{2}(\varepsilon - 127) - \frac{n_i}{2}
	\end{align*}

	and thus $\kappa = 2^{23}\frac{3}{2}(\varepsilon - 127)$.
\end{solution}

\vfill

\pagebreak