Timothy Normand Miller wrote:
> There are no direct solutions to NP-complete problems. 

I did know that. :-)

> And this is
> demonstrated by the fact that everyone's P&R algorithm sucks in some
> way or another.  We can apply knowledge, randomization, hill-climbing,
> and other kinds of search to the problem, but there will always be
> limits to how well we can do.

My point is that AI methods have been used successfully to solve similar
problems. There's a decent chance that similar strategies would work
here, even if the similarity is at a fairly abstract level.

> For FOSS, a major obstacle is that the problem isn't interesting
> enough to enough people, and the results aren't glitzy enough.

Well, maybe. To me, it sounds like the sort of thing a CS or CSE major
could get quite excited about. In fact, if I were 20 years younger and
still in college, I'd probably tackle it. Heck, if somebody paid me a
stipend, I'd work on it now. It sounds like a great learning project in
AI (which I've been self-studying for a few years now, but never really
applied myself to).

Of course, I'd be starting from zilch understanding of this particular
problem.

>>Is the
>>floorplanning algorithm for an FPGA really all that different from the
>>ones used for PCB-layout?
> 
> It's completely different.  Yes, there are superficial similarities
> and certainly some deep ones, but the problems are defined in
> radically different ways.  I went into this for one of the interview
> questions.

Yeah. Of course, what you said made the PCB problem sound *harder*, not
easier.

>>Anyway, this could be at about the level of a university semester
>>project, IMHO.
>
> Pieces of it could be.  A better FPGA P&R algorithm could be a Ph.D.
> dissertation.

PhD candidates do exist. ;-)

For our purposes, of course, "better" is nice but not obligatory. I
don't imagine anyone expects a free synthesis tool (right term?) to beat
a proprietary one on its first outing. We'd just expect it to "work", right?

Cheers,
Terry

-- 
Terry Hancock ([EMAIL PROTECTED])
Anansi Spaceworks http://www.AnansiSpaceworks.com

_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to