Re: gEDA-user: gnucap - oscillator example

[al davis <ad151@xxxxxxxxxxxxxxxx>]

On Tuesday 08 July 2008, Dave N6NZ wrote:
> Schematic driven simulation is practical for homeworks, but
> it doesn't scale.

For homeworks, they should learn how to use a netlist.

Actually, schematics do scale if you do it right.

> Of course, we're talking about SPICE here, and nobody would
> do a SPICE simulation that large, but I still believe the
> scalability issue applies.

We are talking about gnucap here.  Gnucap scales better than 
spice does.  Depending on what models you use and how you set 
the options, it can be anything from more accurate than Spice 
(and slower) to competitive in speed with some of 
the "fast-spice" simulators.  I have some benchmarks that take 
hours to run in NGspice, seconds to run in gnucap.  There is 
more work to do in this area.

The default settings are usually slightly more accurate than 
spice.  That is one of the reasons the oscillator example 
works, and can give meaningful distortion measurements.

One common source of huge netlists is parasitic extraction from 
PC boards or IC's.

> I realize most gEDA users are not doing systems of the scale
> I mentioned above.  But it is all too common for posters on
> this list to dismiss scalability issues just because it
> doesn't apply to what they happen to be working on.

I know!!!!!!!!!!!!  For circuits big enough, run time in NGspice 
usually scales quadratically.  For the same type of circuits, 
gnucap scales linearly.  The theoretical worst case is cubic, 
but I have never seen that in practice.





_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user