That would likely make operations on the default float slower than they need to be. After all, the CPU is probably implementing IEE 754: How do you handle it wh

That would likely make operations on the default float slower than they need to be. After all, the CPU is probably implementing IEE 754: How do you handle it when a NaN bubbles up from below?