CORDIC Trigonometric Algorithm on an HP17BII

[Charles]

There has a great deal of very useful discussion on programming trigonometric functions on financial calculators such as the HP 17BII. Has anyone considered putting a CORDIC algorithm on this calculator or is it just too inefficient compared with other approaches ? I'd try but think that it is beyond my abilities.

[Chris Dean]

Charles

I have found this document on the internet

http://www.fpga-guru.com/files/crdcsrvy.pdf

Is it a good reference or do you have any other reference material that is better?

Chris

[Eric Smith]

The best general reference on algorithms for elementary functions, including both "traditional" and "shift-and-add" algorithms like CORDIC, is "Elementary Functions: Algorithms and Implementation", 2nd Edition, by Jean-Michel Muller.

I'm not very familiar with the capabilities of the 17BII, but implementing a decimal CORDIC generally requires two nested loops, and a table of precomputed constants with basically one table entry per digit of precision.

I suspect that series approximations are far easier to implement on the 17BII than CORDIC would be.

[Gerson W. Barbosa]

Quote:

---

I suspect that series approximations are far easier to implement on the 17BII than CORDIC would be.

---

And a software implementation of CORDIC would not be efficient since it is more suitable for hardware implementation, as you know. Anyway, Charles could try to convert an HP-32Sii/33s program I presented in this thread (Message #4), which is translation of a TI program:

http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/archv016.cgi?read=103989

As of series and polynomial approximations on the HP-17BII, here we have them in six flavors:

http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/articles.cgi?read=695

Regards,

Gerson.

[Eric Smith]

Binary CORDIC is actually pretty efficient on most RISC processors.