What is overflow in floating-point arithmetic?

• A. Underflow to zero
• B. Result exceeds the maximum representable value, often infinity.
• C. Result is truncated to an integer.
• D. Exponent wraps around.

Answer: (B)

Overflow in floating-point arithmetic happens when the result is larger in magnitude than the format can represent. Since floating-point numbers have a limited range, a calculation that would exceed that range cannot be stored as a finite number, so the result becomes infinity (or signals an overflow). The sign of the result matches the sign of the computed value.

This is different from underflow, which is about results that are too close to zero to be represented as normal numbers. Truncation to an integer is about converting a floating-point value to an integer, not about exceeding the representable range. Exponent wrap-around would be a hardware error or undefined behavior, not the standard behavior of overflow.