[net.micro.68k] Wanted: trig fcns in C or for 68k

geof@imagen.UUCP (Geoffrey Cooper) (01/10/86)

I am in need of an implementation of floating point routines for the 68000.
Actually, the greatest need is for just the trig functions, following IEEE
floating point (I have the rest, but the trig functions are implemented
poorly).  They could be implemented in assembly language or C.

- Geof Cooper
  Imagen
  {decwrl,saber}!imagen!geof
  imagen!geof@decwrl

tgl@a.sei.cmu.edu (Tom Lane) (01/15/86)

In <181@imagen.UUCP> geof writes:
> I am in need of an implementation of floating point routines for the 68000.
> Actually, the greatest need is for just the trig functions, following IEEE
> floating point (I have the rest, but the trig functions are implemented
> poorly).  They could be implemented in assembly language or C.

If you're willing to write the code yourself, there is a truly
outstanding book called "Software Manual for the Elementary Functions",
by Wm Cody & Wm Waite, Prentice-Hall, 1980.  It gives flowcharts,
coefficient values, and extensive discussion for implementation of
first-rate routines for the basic transcendental functions.
You can write code in assembly, C, or whatever given the info in
the book; you do need a pre-existing set of basic arithmetic routines
(+,-,*,/) which should be IEEE quality in order to get good results.

		tom lane (lane@a.cs.cmu.edu on ARPA)

spellman@hplabsb.UUCP (Miles Spellman) (01/16/86)

motorola sells a package that has double, single and
"fast floating point" formats with a good selection of trancendentals.
Their fast floating point is a format optimized for the 68000 and
does 32 bit math about twice as fast as the IEEE 32 math equivalent.
It was called " 68343 FAST FLOATING-POINT ", good stuff!

			Miles Spellman
			hplabs!spellman