bond@chalmers.UUCP (10/29/87)
In LightSpeed Pascal V1.11a there is no support for the floating point
processor MC68881 in the Mac II. This is to announce that I have made a
small library supporting MC68881 and callable from LSP. I am prepared
to distribute it freely to the net if there is any interest. It right
now supports arithmetic with reals (i e no doubles) only.
I decided to make the library after comparing execution times for part of
an application as indicated in the following table. I couldn't accept
to have to write programs in FORTRAN once again.
Absoft FORTRAN/68020 5 Tickcounts
LSP V1.11a 210 -"-
LSP V1.11a+Library 15
The day after the library was completed I received TML Pascal 2.5 and got the
result
TML Pascal 2.5 38 Tickcounts
However the problem is very unfavorable to the TML since their library only
supports the type extended.
To add two reals you write e.g.
c:=FADD(a,b);
This is of course not the most proper way; the same metod is used in TML Pascal.
Absoft FORTRAN supports the notation c=a+b.
I guess that other people have written similar libraries and perhaps there is
one which is more complete.
I have included a function
FUNCTION MandelIter(z,c:complex;max:integer):integer;
which performs a number of iterations of the form
z:=z*z+c
Let me also mention that I sometimes experience problems when using userwritten
libraries in the LSP environment, although there are no problems when the
same program is run as a separate application.
Magnus Bondesson
Dep of Computer Science
Chalmers University of Technology
S-412 96 G|teborg
SWEDEN
UUCP: ..!mcvax!enea!cs.chalmers.se!bondir353@sdcc6.ucsd.EDU (Matthew Grayson) (10/31/87)
In article <2075@chalmers.UUCP> bond@chalmers.UUCP (Magnus Bondesson) writes: > >In LightSpeed Pascal V1.11a there is no support for the floating point >processor MC68881 in the Mac II. This is to announce that I have made a >small library supporting MC68881 and callable from LSP. I am prepared >to distribute it freely to the net if there is any interest. YESYESYESYES PLEASE, OH PLEASE PLEASE PLEASE P O S T I T ! ! ! ******************** Better yet, mail a copy to me NOW, and I'll be your friend for life. I, too, can't stand Fortran, Love LSP, and must do alot of floating point. Well, I could go on being effusive, but you get the point. I doubt that I'm alone, and a lot of Mac II owners who do light numerical analysis will benefit tremendously from your work. I'm posting rather than emailing in order to drum up support from the net. Thanks ever so much, Matt Grayson (sdcc6!ir353@sdcsvax.ucsd.edu)