diff --git a/libs/util/include/psemek/util/fixed_point.hpp b/libs/util/include/psemek/util/fixed_point.hpp
index 20295b43..8827092a 100644
--- a/libs/util/include/psemek/util/fixed_point.hpp
+++ b/libs/util/include/psemek/util/fixed_point.hpp
@@ -20,7 +20,7 @@ namespace psemek::util
 	 *  represented as an (I+F)-bit-wide standard signed integer type, and the
 	 *  stored value is interpreted as representing the fraction (value / 2^F).
 	 *  It is capable of representing values in the range from (- 2^(I+F-1)) / 2^F
-	 *  to (2^(I+F-1) - 1) / 2^F, with a fixed precision of 2^F.
+	 *  to (2^(I+F-1) - 1) / 2^F, with a fixed precision of 1 / 2^F.
 	 *
 	 *  For example,
 	 *
@@ -40,7 +40,9 @@ namespace psemek::util
 	 *  separate question...).
 	 *
 	 *  Multiplication and division are implemented by casting to a 2x wider integer type,
-	 *  to increase precision and prevent overflows.
+	 *  to increase precision and prevent overflows. Note that the result is still truncated
+	 *  to the original size, i.e. multiplying two 16.16 fixed-points gives a 16.16 fixed-point
+	 *  as well.
 	 *
 	 *  Typical usage of this class would be to make a type alias like
 	 *
@@ -345,6 +347,8 @@ namespace psemek::util
 		if (p < 0)
 			return fixed_point<I, F, S>(1) / pow(x, -p);
 
+		// Binary exponentiation algorithm
+
 		fixed_point<I, F, S> result(1);
 		while (p > 0)
 		{