Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

what is the difference of the three accuracy in nanosim?

Status
Not open for further replies.

lhlbluesky

Banned
Advanced Member level 2
Joined
Mar 30, 2007
Messages
558
Helped
5
Reputation
10
Reaction score
5
Trophy points
1,298
Location
china
Activity points
0
nanosim vs spectre

in nanosim, we can set the accuracy for our circuit. there are three accuracy: simulation accuracy, model accuracy, netlisting accuracy. i want to ask, what is the difference of the three accuracy? and how to understand them?

the default accuracy is 2, if i want to make the nanosim accuracy equivalent to spectre accuracy(moderate etc), which accuracy should i use? if i set accuracy=3, is it ok? and if i want to set nanosim accuracy=3, which of the following is right:
1, simulation accuracy=3
2, model accuracy=3
3, netlisting accuracy=3
4, simulation accuracy=3, model accuracy=3 and netlisting accuracy=3
besides, my circuit has 9000 transistors or so.

i expect your advice. thanks all.
 

Re: nanosim vs spectre

lhlbluesky said:
any ideas, pls?
See **broken link removed**, especially santhoshv78' contribution from Mon, 06 Sep 2004 8:17 . Don't know if it can help.
 
Last edited by a moderator:

nanosim vs spectre

as indicated in another thread pointed by erikl ..

nanosim is a fast spice meant to be used for big circuits that can't simulated in spectre/hspice/eldo..

set_sim_eou is a macro command that internally controlled more than 50 different commands..

there are lot of things that vary with this sim_eou command ..
e.g. internal time/voltage/current resolution , tolerances
whether to use complete mos equations or to use use average mos model or use tables instead of equations
how to deal with caps , resistors in desing ,,,
there are so many things that are controlled internally...

higher level of set_sim_eou better the accuracy and more runtime...

normally level 3 is used for functional verification for soc's ..
if you have a analog ckt go for 4,5 ...

normally fast spice like nanosim needs some iteration to get the right settings to be used

eou level requirement changes according to technology nodes as well as circuit toplogy ...
 

Hi,

Just by transistor count, its impossible to say which accuracy will work out well. It is one of the criteria but not at all enough. You surely need a lot of tries to arrive at a proper value of EOU. You must know what are the different modules and which need more accuracy too. u can set EOU accuracy module wise as well.

We start with 2,2,2 and keep increasing this depending on various factors. If highly sensitive analog portions are present, better go for 4 4 for model and sim. Keep model and sim acc as close as possible.

Try not to incraste NET accuracy to beyong 2. Let the netlister do its approximations. They usually do not have a huge impact on results but have good performance impact. So its good to keep NET acc to 2. Play with model and sim acc.

Hope it helps.

Kartik

Added after 3 minutes:

Hi,

By the way, since the count is 9000, which is very very low compared to Mixed Signal SOCs coming these days, I would suggest you can start with 4 for model and 3 for sim and 2 for net.

For some sensitive analog modules, selective make them 4,4,2. It should work perfectly.

Regards,
Kartik
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top