This section discusses how the Alpha architecture handles denormal, plus and minus infinity, and NaN values, and provides suggestions for maximizing floating-point performance with these types of values.