Use a double and accept some loss of precision, or use a bignum library like GMP: The GNU MP Bignum Library