Quick Fix Multiplication Library - almost complete!

[kcr103@psuvm.psu.edu (Kenneth C. Richardson)]

A few months back, I asked what high speed integer routines programmers
would want.  I am almost finished with all of the following routines:

QKMul2    Unsigned 16 x 16 --> 16 bit result
QKsMul2   Signed   16 x 16 --> 16 bit result
QKMul4    Unsigned 32 x 32 --> 32 bit result
QKsMul4   Signed   32 x 32 --> 32 bit result
QKFixMul  32 x 32 bit fixed point multiply (16 bit fraction)
QKsMul48  Signed   32 x 32 --> 64 bit result

These routines are 2-3 faster than their ORCA counter parts.  They also
do not use lookup tables and do not require two banks of RAM to
function.
The largest routine is about 5k.  I am planning to sell the math
library
for $20.00 to anyone who is interested.  Ken

send email to kcr103@psuvm.psu.edu