|
|
Magic Mailing List |
|
From: cfk (cfk AT pacbell DOT net) Date: Sat May 03 2003 - 14:01:29 EDT
Dear John & Stefan:
Near as I can tell, the octtools are a huge set of source that will take
some time to understand. Before undertaking that, I guess the first question
is how do they relate to magic, and to the extent they do relate to magic,
how can we get one or more of these tools compiled under the tcl interpreter
so they can play nice with magic, tclspice and others. Certainly, adding
place & route to magic is potentially a huge undertaking. My original
intention in bringing it up yesterday was to get some momentum with the
group to examine the utility of adding a little bit of functionality
relating to place and route into magic.
If that should be done with a program that is part of octtools, that
would be a wonderful undertaking.
Charles
----- Original Message -----
From: "john wood" <john.wood AT multigig DOT com>
To: "cfk" <cfk AT pacbell DOT net>
Cc: "Stefan Jones" <stefan.jones AT multigig DOT com>
Sent: Saturday, May 03, 2003 4:28 AM
Subject: Re: EDIF
> Charles,
>
> For what they are worth,
> attached are my own notes I prepared
> as I was playing with octtools and
> reading the available documentation.
>
> I only got so far as to conclude it
> was a good system in which to experiment
> with new ideas, but no good for
> commercial chip design because of
> capacitity restrictions, speed
> (simulated annealing is SLOW}
> and 2-metal only routing with
> channels.
>
> Looking at Itools, they dont seem
> to be that much better and we are going
> to have to buy a real P&R tool soon.
>
> John
>
>
> On Fri, 2003-05-02 at 19:08, cfk wrote:
> > Dear John:
> > I did get the octtools zip file, configured and installed it without
> > errors. Can you point me to some tutorial somewhere that will allow me
to
> > appreciate the care and feeding of these tools. I only familiar with
them by
> > name and have not used any of them before. I would like to study their
use
> > before making any more comments on this group about place and route with
> > Magic.
> >
> > Charles
> >
> > ----- Original Message -----
> > From: "john wood" <john.wood AT multigig DOT com>
> > To: "cfk" <cfk AT pacbell DOT net>
> > Sent: Friday, May 02, 2003 10:09 AM
> > Subject: Re: EDIF
> >
> >
> > > Charles,
> > >
> > > Stefan here has fair bit of work on this
> > > but using timberwolf (within octtools )
> > > for place / route then
> > > using oct2mag convertor.
> > >
> > > Stefan is back in on Monday and I am sure
> > > he will help you.
> > >
> > > In the meantime have a look at
> > > http://sourceforge.net/projects/octtools/
> > > we got so far as running it under linux
> > > and came up with some cheats for doing
> > > 'over-the-cell' routing also.
> > >
> > > Recently we have been testing itools
> > > (www.twolf.com) which is the commercial
> > > version of timberwolf place+route.
> > >
> > > best regards,
> > >
> > > John
> > >
> > >
> > > On Fri, 2003-05-02 at 17:33, cfk wrote:
> > > > I wonder if we can discuss EDIF or perhaps the more general topic of
> > getting
> > > > a portion of a digital design expressed in Verilog or VHDL into
Magic
> > for a
> > > > while. Is there a way to get a netlist of a digital design into
Magic
> > > > somehow. I can see the notion of creating a netlist by hand, but I
that
> > will
> > > > soon become too cumbersome.
> > > >
> > > > If not, what is the most appropriate way to add the capability to
read
> > an
> > > > EDIF or EDIF like netlist into Magic for the purpose of then placing
and
> > > > routing by hand INV, OR, AND cells and using its route command to
route
> > > > those cells.
> > > >
> > > > I know this pandoras box has many components, I am just trying to
pick a
> > > > place to start that has a reasonable probability of getting as far
as
> > > > possible with the least amount of work on the part of all concerned.
> > > >
> > > > Charles
> > > >
> > > >
> > > >
> > > >
> > >
> >
> >
> >
>
>
|
|
|
|