[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Christmas Gift for Programmers



The Quickfix Multiplication Library has been available for
several weeks. If you want to surprise your favorite Apple ][gs
programmer with a new toy, this new library might do it.  A description
of the library follows.


QuickFix is a library of high speed multiplication routines that
accelerate integer multiplication routines three fold over standard
multiplication routines.  Moreover, because they do not use a lookup
table, they do not eat up two banks of RAM and thus can easily fit
into the cache of today's hardware accelerators.  (The QkMul2 routine
uses about 5k of space.)  The library contains the following routines:

QkMul2   - unsigned 16x16 bit multiply with 16 bit result
QksMul2  - signed 16x16 bit multiply with 16 bit result
QkMul4   - unsigned 32x32 bit multiply with 32 bit result
QksMul4  - signed 32x32 bit multiply with 32 bit result
QkFixMul - 32 bit fixed point multiply with 16 bits of fraction

Currently, the routines are in a library file usable by ORCA/M.
One simply has to copy the library file into the libraries folder
of your ORCA program folder.  The documentation explains how to
use the routines from assembly.  There is also a brief description
for using the routines from "C".  There is presently no support for
Pascal, but I suspect someone out there knows how to do it.

I have found these routines indispensible for math intensive programs.
I first used a prototype math routine to speed up a graphics algorithm
from 2.5 minutes to 40 seconds.  The graphics program became usable for
the first time, because one did not have to wait forever. Imagine if
you have a 3D rendering program that now will require half the time
because of the QkFixMul routine.   When I realized how much faster any
program could become with these math routines, I knew I had to make
them
available to everyone.

The QuickFix routines were among the most difficult of my programming
efforts.  They required over 10,000 lines of assembly code and I am
planning even more routines for the future.  If you need a specific
math routine, please let me know!

The introductory price is $24.95 which includes a license to use the
routines in any personal programs or shareware programs.  A license
for commercial software is also available (also quite reasonable).
Shipping is included in the above price.

Payments are by Check or Money Order to:

Kenneth C. Richardson
829 N. Second Street, 1st Floor
Philadelphia, PA 19123
(215) 923-2355

email: kcr103@email.psu.edu