help-gnucap
[Top][All Lists]
Advanced

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

Re: [Help-gnucap] Multiple instances having the same instance name


From: dev
Subject: Re: [Help-gnucap] Multiple instances having the same instance name
Date: Mon, 15 Jun 2015 16:12:34 -0400
User-agent: Roundcube Webmail/1.1.1

 

Thanks Al, for your informative reply. 

ngspice does not compile this and I have not tried any other program
yet. 

Since it simulates the two instances correctly i.e. internally the two
instances are truly distinct by design (and not just by happenstance in
my specific case), it is not an issue. 

Given the implementation effort and complexity you have mentioned, I can
just be careful with the instance-names and all is OK. 

Thanks again. 

Brunno. 

On 2015-06-15 14:49, al davis wrote: 

> On Monday 15 June 2015, address@hidden wrote: 
> 
>> That is the instance name XD is used twice for two different
>> instances. The whole circuit works well despite what I think
>> is an error.
>> 
>> Is this intended behaviour or should this fail to compile?
> 
> Intended behaviour.
> 
> Just curious .. what do others do?
> 
> I am not sure from a user perspective which way is preferred. 
> My gut feel is that it doesn't matter.
> 
> From a performance perspective, checking by any method is extra 
> work which would make the total run time slower.
> 
> Old versions of gnucap had an option "dupcheck" to turn checking 
> on and off, defaulting to off, but that got removed as being in-
> the-way and incorrectly implemented in the conversion to 
> plugins.
> 
> Technically, to do it efficiently involves using hash tables, 
> which may happen eventually for other reasons. Any other 
> method would be too slow for large circuits. Using a linear 
> search (O(n^2)) would be so slow that it would take more time to 
> check for dups than the whole rest of the simulation combined. 
> Even a binary search (O(n*log(n))) would have a significant 
> impact on the batch runtime of a large circuit.
 


reply via email to

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