This is the primary reason (that I can see) that a built-in software floating point scheme would be useful (like that mentioned in feature request #40). Troy