Reading Assignment and Discussion Topics
Computer Science 471

for class on Tuesday Oct. 1, 1996

Please read Sections 4.6 through 4.12 of the Patterson & Hennessy text, and be prepared to discuss the following issues:

1. What would a floating-point multiply circuit look like? Let's make one in the style of Figure 4.42. Figure 4.43 purports to describe floating-point multiplication, but is instead just a copy of Figure 4.42. The correct version of Figure 4.43 can be found at http://www.cs.princeton.edu/courses/cs471/fig4-43.gif. An integer-style multiplier will be needed somewhere in the middle of our circuit; we will not work on the insides of this. Single precision, IEEE format. Your instructor is always a little shakey on rounding and overflow; any help you can give him with this will be appreciated.

2. Famous Princeton guy John von Neumann is quoted on pages 293-4 of the text expressing arguments for and against floating point. A page of the 50-year-old document in which these quotes appear is reproduced on http://www.cs.princeton.edu/courses/cs471/vonN.gif; the relevant part is Section 5.3. What do you think of von Neumann's original arguments and why did he decide this issue the way he did? (And what are these ``scale factors,'' anyway?) Was he wrong? Was he wrong then?