[net.micro.amiga] Amiga Floating Point

bobp@amiga.UUCP (Robert S. Pariseau) (11/01/85)

[To Curt Jutzi -- I, personally, have TRIED to reply to you using the
golem!jutz path in your earlier news.  The responses have been bounced
back from tektronix with a message saying your machine was following
an "illegal protocol".  I hope this, at least, gets through.]

Floating point support exists on the Amiga in several forms for several
purposes.  The floating point which is directly supported by expression
evaluation in the V1.0 release Lattice C for the Amiga is a 64 bit
software implementation of the IEEE format.  As such, it is *>SLOW<*.

The floating point in ABasiC is a 32 bit IEEE format running in a
somewhat simpler exception handling environment.

Motorola Fast Floating Point format is available (in 32 bits only) in
two standard libraries.  The fundamental math routines are in the
library mathffp which resides in the Writeable Control Store.  The
corresponding transcendental functions are in the library mathtrans
which is disk-based (i.e., it's found in the LIBS: directory and is brought
into memory and shared between programs as needed).

To access routines in these libraries you do an OpenLibrary() call
and then call the desired routine using one of the interface entrypoints
defined in amiga.lib.

I'll post a working example on the net soon.

Yes, we are working with Lattice to make the 32 bit IEEE format and
the 32 bit Motorola format routines directly available from the language.

--------
We do not expect to make the Amiga schematics public anytime soon.  There
will, however, be more info available about the timing and characteristics
of the expansion bus.  Curt, if you have questions about the hardware,
I'll try to get them answered.  Please be sure to include a phone number
so that I don't have to post the replies to the group.

pgydow@sunybcs.UUCP (Bruce Dow) (11/15/85)

             eat your heart out line eater

 I would be very interested in the hardware info.
whether or not i get an amiga or an atari st or stt
depends on what hardware info i can get  
 ie. the more info the more useful the machine..
   subject to the usual non-disclosure/copyright/patent
ect...
                                        John Dooley


                                        ( I disclaim everything
                                          (ala DEC ))