Re: [f-cpu] DivMod unit

[Cedric BAIL <cedric.bail@free.fr>]

> > Soustraction are the same as adder so there is the same type of
> > algorithme.
> 
> As noted before, there are a few boolean optimisations to do on a
> classical adder : tying the carry in to 1 can be propagated in the
> equations, as well as the negation of the input.
> 
> As i also wrote, there is a very interesting document about
> arithmetic operators at
> http://www.iis.ee.ethz.ch/~zimmi/publications/comp_arith_notes.ps.gz
> (thank you michael !)
> 
> WHYGEE
So you have now a new version, that will correctly work in 8, 16, 32 and 64 bits
mod. I add a first release of what the eu_idiv unit will look like. The code
is now clear and human readable. I remove all debugging line, and only the
necessary code is present.

The test best say some stupid thing when you are in mod > 16, I think that's my
error but i don't know were I do it. (The modulo always say 0) If somebody
know why simili say that...

And what did you think about the pipelining of the 8bits part ? I know the
core will perhaps not be more orthogonal.

A+
  Cedric

eu_idiv.tar.bz2