On the precision attainable with various floating-point number systems | IEEE Conference Publication | IEEE Xplore