[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnucap-devel] method and error (was: transient -- "trace" default)

From: al davis
Subject: Re: [Gnucap-devel] method and error (was: transient -- "trace" default)
Date: Mon, 13 Jan 2014 01:45:00 -0500
User-agent: KMail/1.13.7 (Linux/3.2.0-4-amd64; KDE/4.8.4; x86_64; ; )

On Saturday 11 January 2014, Felix Salfelder wrote:
> > On the surface we only care about h small.  But really that
> > is  not so.  There are times when we want to allow a large
> > time step and HIDE the error with the intent of ignoring
> > it.
> this hasnt occurred to me yet. i would have tried to replace
> the netlist with a smaller/faster one.

If you are just doing small experimental circuits you can do 
that.  For large circuits, probably not.  The small caps we wish 
to ignore are mostly parasitic caps in device models.  Some of 
the large circuits are automatically generated from layouts.

> the pulse patch makes things worse for me. the number of
> iterations increases from ~3000 to ~5000. i think that is
> because the patch moves the events from the discontinuities,
> where my mechanism requires them.

If you want exact  timing, use the event queue, not the 
"ambiguous_event" mechanism.

Probably bm_pulse should use the event queue instead.

When there is a discontinuity, you need an event just before and 
just after.  With the event queue, the whole list can be built 
in advance.  With the ambiguous_event mechanism, it's one step 
at a time.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]