\myheading{Exponentiation}

Using equation \ref{eq:1} we may quickly notice that

\begin{equation}
b^n = \underbrace{b \times \cdots \times b}_n = base^{(\log_{base} (b))*n}
\end{equation}

That works with any logarithmic base.
In fact, this is the way how exponentiation is computed on computer.
x86 CPU and x87 FPU has no special instruction for it.

This is the way how pow() function works in Glibc: \url{https://github.com/lattera/glibc/blob/master/sysdeps/x86_64/fpu/e_powl.S\#L189}:\\

\begin{lstlisting}[caption={Glibc source code, fragment of the pow() function}]

...

7:	fyl2x			// log2(x) : y
8:	fmul	%st(1)		// y*log2(x) : y
	fst	%st(1)		// y*log2(x) : y*log2(x)
	frndint			// int(y*log2(x)) : y*log2(x)
	fsubr	%st, %st(1)	// int(y*log2(x)) : fract(y*log2(x))
	fxch			// fract(y*log2(x)) : int(y*log2(x))
	f2xm1			// 2^fract(y*log2(x))-1 : int(y*log2(x))
	faddl	MO(one)		// 2^fract(y*log2(x)) : int(y*log2(x))
	fscale			// 2^fract(y*log2(x))*2^int(y*log2(x)) : int(y*log2(x))
	fstp	%st(1)		// 2^fract(y*log2(x))*2^int(y*log2(x))

...

\end{lstlisting}

x87 FPU has the following instructions used Glibc's version of pow() function:
FYL2X (compute $y \cdot log_2 x$), F2XM1 (compute $2^x–1$).
Even more than that, FYL2X instruction doesn't compute binary logarithm alone, it also performs multiplication operation, 
to provide more easiness in exponentiation computation.

It works because calculating $2^x$ (exponentiation with base 2) is faster than exponentiation of arbitrary number.

Using hacker's tricks, it's also possible to take advantage of the IEEE 754 format and SSE instructions set:\\
\url{http://stackoverflow.com/a/6486630/4540328}.