diff --git a/ex06/crossSwitchPlot.png b/ex06/crossSwitchPlot.png
new file mode 100644
index 0000000..2dab5e5
--- /dev/null
+++ b/ex06/crossSwitchPlot.png
Binary files differ
diff --git a/ex06/kn06.pdf b/ex06/kn06.pdf
index 4ce2993..1481f15 100644
--- a/ex06/kn06.pdf
+++ b/ex06/kn06.pdf
Binary files differ
diff --git a/ex06/kn06.tex b/ex06/kn06.tex
index e31bba1..3b29deb 100644
--- a/ex06/kn06.tex
+++ b/ex06/kn06.tex
@@ -120,12 +120,15 @@
     \begin{enumerate}
         \item There are $10^{2i}$ crosspoints (in $\times$ out)
         \item At maximum there are $10^i$ crosspoints active simultaneously ($\min$(in,out))
-        \item %TODO
+        \item $C_N$ has the double slope of $X_N$, the ratio has a negative slope since $C_N$ is growing faster than $X_N$. The ratio describes kind of the efficiency of the switch, we see that it decreases for a large number of connections.\\
+        \includegraphics[width=.5\textwidth]{crossSwitchPlot.png}
         \item $n$ for minimal number of crosspoints: $\left\lceil \sqrt{\frac{10^i}{2}}\right\rceil=3,8,23,71,224$\\
         $k\geq 2n-1$ : $k=5,15,45,141,447$
         \item We use the minimal number of crosspoints above:\\
         $C_{N,opt}=\frac{N}{n}\cdot n\cdot k + k\cdot \frac{N}{n}\cdot\frac{N}{n}+ \frac{N}{n}\cdot k\cdot n\\
         =153,5337,175043,5617042,178486160$
-        \item %TODO
+        \item The optimal number of crosspoints for large N is lower than the number of crosspoints in the crossswitch. For small $N$ vice versa.\\
+        $C_{N,opt}$ is numerically not stable because of the ceiling function for $n$.\\
+        \includegraphics[width=.5\textwidth]{multiStageSwitchPlot.png}
     \end{enumerate}
 \end{document}
diff --git a/ex06/multiStageSwitchPlot.png b/ex06/multiStageSwitchPlot.png
new file mode 100644
index 0000000..b456485
--- /dev/null
+++ b/ex06/multiStageSwitchPlot.png
Binary files differ
diff --git a/ex06/plots.py b/ex06/plots.py
new file mode 100755
index 0000000..5341289
--- /dev/null
+++ b/ex06/plots.py
@@ -0,0 +1,37 @@
+#!/usr/bin/python
+import matplotlib.pyplot as plt
+import numpy as np
+
+def copt(N):
+    n=np.ceil(np.sqrt(N/2.))
+    k=2*n-1
+    return 2*N*k + k*(N*1./n)**2
+
+
+fig = plt.figure()
+ax = fig.add_subplot(1,1,1)
+N=np.arange(1,10**5,1)
+ax.plot(N, N**2,label="C_N")
+ax.plot(N,N,label="X_N")
+ax.plot(N,1./N,label="X_N/C_N")
+ax.set_xscale('log')
+ax.set_yscale('log')
+#ax.axis((1,10**5,1/100.,10**10))
+ax.legend()
+fig.show()
+
+raw_input('press return to continue')
+
+fig = plt.figure()
+ax = fig.add_subplot(1,1,1)
+N=np.arange(1,10**5,1)
+ax.plot(N, N**2,label="C_N")
+ax.plot(N,copt(N),label="C_N,opt")
+ax.plot(N,N**2*1./copt(N),label="C_N/C_N,opt")
+ax.set_xscale('log')
+ax.set_yscale('log')
+#ax.axis((1,10**5,1/100.,10**10))
+ax.legend()
+fig.show()
+
+raw_input('press return to continue')