This time, trying to multiply 471 by a floating point number greater than
one.  This looks so immensely bogus that I think I have to chalk it up to
bum hardware.  Does anyone have a good guess as to whether the motherboard
can convince a 486 to take random FP exceptions, or is it guaranteed to be
the chip that's toast?  (I've been looking for an excuse to get a new MB,
anyway...)