Re: [ng-spice-devel] Hello World


To ng-spice-devel@ieee.ing.uniroma1.it
From Paolo Nenzi <pnenzi@ieee.ing.uniroma1.it>
Date Thu, 9 Mar 2000 14:27:04 +0100 (CET)
Delivered-To mailing list ng-spice-devel@ieee.ing.uniroma1.it
In-Reply-To <38C70B6C.A6E830F2@galaxy.nsc.com >
Mailing-List contact ng-spice-devel-help@ieee.ing.uniroma1.it; run by ezmlm
Reply-To ng-spice-devel@ieee.ing.uniroma1.it



On Wed, 8 Mar 2000, Mathew Sienko wrote:

> Hi Everyone,

>    As it stands now, the acquisition of tools is a major barrier for
> engineering startups.  I have some friends who are in small design
> startup companies and they brag about the fact that they have a cadence
> liscense.  This is NOT cheap.  Untill this situation changes, only large
> companies will be able to do IC design; those that are already large,
> and the venture capitol firms that pay for the little ones to buy their
> tools.  I dream of the day where the huge costs don't really start
> untill you send a design out to fab.

I think that we are still very far from this point. Candence, but I think
other companies do something similar, have a sort of "foundry
certification program" that assure their customers of the correspondence
between their simulation and the final product (the IC). I hope that one
day each one of us can program it's bios, modify its motherboard chipset,
or even build his own "compatible" microprocessor.
  

> I realize that this will NOT happen overnight, but I
> feel energized by the knowledge that there are people working on a
> professional quality simulator.
There exists a lot of good EDA projects in the open source community and
their "goodness" lower as they try to get to the mask level. We have a lot
of VHDL or Verilog systems some of them very usable others sufficient for
most needs, but when we try to reach the spice level or worse the layout
lever, the number and quality of software rapidly derease. I think that
this depends on the "relative interest" of the community. Many people are
interested in VHDL simulation and far less in Design Rule Check or mask
fabrication.

> large scale.  I've done a little bit of reasearch into specter (of
> course, they don't let you see the code) and would like to make ngspice
> able to have the same or better performance.   
Thne you are in the right place. As other open source projects we want to
get the best from each simulator.


> An other thing I think
> would be good is the ability for this to run on, and take advantage of,
> MP alphas.  This may become moot by the time this happens, but alphas
> have been about twice as fast as the competition in fp calculations for
> a few years now.

We are currently in a stall situation where, I think, other people are
waiting the completition of my code, a unit that will handle all the data
generated by the simulation. This small unit will be the first of the new
code, the GPL'ed one, (Sorry for the long time...) and will be probably be
multithreaded.

In the meantime we have "patched" and "modified" the existing spice code,
the one in the experimental directory of the ftp tree. This is a "dead"
code, we want build a GPLed simulator, so if the license of spice do not
change, we have to recode it from scratch.  

>    Well, I just downloaded the code and will start looking at it to see
> what sense of it I can make.  I'm familiar with C, but am not really a
> programmer.  I am coming at this from the viewpoint of an analog
> designer and want to help make this a professional quality simulator.

Let us know your advancemets,

welcome aboard!

Paolo Nenzi


Partial thread listing: