Re: New and faster algorithm for multiplication

In article <HYyTe.5224$qE.1169977@xxxxxxxxxxxxxxx> hselasky@xxxxxxx writes:
> *** T. Winter wrote:
> > In article <zXlTe.5189$qE.1164274@xxxxxxxxxxxxxxx> hselasky@xxxxxxx
> > writes:
> > > Hi,
> > >
> > > I think I have found a new and faster algorithm for multiplication.
> >
> > i think the following is faster:
> >
> > static unsigned mult(unsigned x, unsigned y, unsigned z) {
....
> Can you explain what "x", "y" and "z" are, and what "mult()" computes. I got
> that your "mult(x,y,z)" is not equal to "((x*y)+z)".

For what x, y and z?

> But I think I get what
> your circuit is about, and mine is twice as fast, because I have that last
> full adder optimized into my circuit.

It is a serialisation of the standard Wallace tree implementation for
multiplication.

But a more appropriate newsgroup for this is comp.arch.arithmetic.
--
*** t. winter, cwi, kruislaan 413, 1098 sj amsterdam, nederland, +31205924131
home: bovenover 215, 1025 jn amsterdam, nederland; http://www.cwi.nl/~***/
.

Quantcast