There is probably nothing so useless as a technical textbook author who thinks his readers are so smart that the practical details will bore them. Now, there are readers for whom that’s true, but it’s generally just laziness excused as a “problem of knowledge” when it’s not. For example, a computer architecture textbook that refers you to the IEEE 754 standard algorithm for floating point division, but doesn’t explain it, isn’t much use to anyone. A manual that tells you to RTFM is missing the point in a big way.
Unfortunately these sorts of textbooks are preferred at pReStIgIoUs universities because they’re more fun to read if you already know the subject well, so the professors read them and think “wow, those were some insightful comments on the IEEE 754 standard algorithm for floating point division, which everyone obviously knows already”. Then overcoming the uselessness of the textbooks (analogous to the uselessness of the professor’s lectures themselves) becomes another part of the unspoken IQ test which forms the gold-in, gold-out gatekeeping function that keeps these universities prestigious. Which is to say that such classes are selection tests for people who would have figured out the subject on their own, not training for the next rank down of people who could learn it if they were taught properly.
This is why I’ve observed that technical subjects are taught best at community colleges, then at trash universities (like my alma mater), then at state schools, and worst at prestige schools. That is, the quality of instruction is negatively correlated with prestige, the main reason being that prestige is used as a low-key proxy for IQ and conscientiousness, which are character qualities we pretend don’t exist to soothe the status anxiety of women and minorities.
Nah. Or it assumes you have the knowledge anyway, or have dabbled enough.
I view these things as elaborate man pages. They are good for when a highly technical problem arises.
I think the author by this indicates that the details of fp division are considered peripheral to the book. It probably gets far too involved “but if you really want to know, read IEEE 754″.
My main retention was that X/Y = X * (1/Y) and the reciprocal 1/Y can be computed ‘more easily’. So go that route.
See also a couple of (good) advanced intros on the topic below. But honestly, it’s a pretty specialized topic. Are there even numerical analysts anymore?
David Goldberg, Computer Arithmetic, Appendix A, Hennessy & Patterson 1st ed. (Also appears in at least some subsequent editions.)
David Goldberg, What Every Computer Scientist Should Know About Floating-Point Arithmetic
Btw. wikipedia has a pretty good overview of division. There are, of course, many algorithms.