That was at
this topic, and as it developed, someone brought up that the download was probably illegal. I called the publisher listed, and they said, "That's not
our book." Then I emailed Jack Crenshaw himself and he didn't even bother to answer. Apparently someone with an
ACM Digital Library membership downloaded it and then put it on bit torrent, and then where I originally got it proved to be spreading a virus. (I found out about the virus through a web search. Fortunately I was not vulnerable to it since I don't have any Microsoft software or OS on my PC.) I wanted the book on paper anyway, so rather than print the file, I was pleased that our computer-science daughter-in-law found it for me, used, for $8, when some places were asking over $1000 for this out-of-print book! It seems to be mostly a collection of articles the author had in Embedded Systems Programming magazine years ago, which I kept.
If you need the functions fast though, in 16-bit scaled integer which works for most things people think you have to have floating-point for, I have the article and EPROM tables at
http://wilsonminesco.com/16bitMathTables/index.html which will give you the values
hundreds of times as fast as actually having to calculate them, and they will be
accurate to all 16 bits. There are tables for trig and log functions, square root, squares (to speed up multiplies of up to 16-bit by 16-bit numbers, with 32-bit output), multiplication (for when both inputs are 8-bit and the output is up to 16-bit), inverses with 32-bit results (for division, since multiplying by the inverse is much faster than actually dividing), and the bit-reversal tables for fast fourier transforms. The files' usage, contents, and formation methods are described there, and there's sample 6502 & 65816 code for using them.