|Re: HP 35 ROM V2 "exp(ln(2.02) = 2", explanation found|
Message #4 Posted by Eric Smith on 16 Mar 2006, 4:58 a.m.,
in response to message #3 by Jacques Laporte
That's fine, no rush. I see that one of the 35 listings on your site had the copyright notice, and the other (the one you modified?) did not.
I've spent more time studying your site since I posted earlier, and the more I look over it, the more impressed I am. You've done a fantastic job of explaining the algorithms.
The algorithms only changed slightly from the 35 through the early Woodstock series (21/22/25/25C). The first big change to the math routines occurred in the late Woodstock era, in the 67/97, 92, 19C/29C, and maybe the 27. An overview of some of the changes is given in the November 1976 HP Journal article "The New Accuracy: Making 2^3=8" by Dennis Harms.
There were some lesser changes in the 30 series, after which the math code remained fairly stable from that point through the 41 and the Voyager series.
I haven't had time to do any thorough analysis of the differences.
AFAIK, all of the Saturn-based calculators use essentially the same floating-point math routines, derived from the HP-71B. Some of the algorithms used are significantly different from the earlier calculators.