The Designer's Guide Community
Forum
Welcome, Guest. Please Login or Register. Please follow the Forum guidelines.
May 5th, 2024, 3:43am
Pages: 1
Send Topic Print
problem with netlist recreation (Read 11055 times)
alirezad
Junior Member
**
Offline



Posts: 11

problem with netlist recreation
Feb 17th, 2010, 3:40pm
 
Hi,

I am running a simulation with Cadence Spectre and sometimes, even when I click on the "Netlist and Run" button, the netlist is not updated.

I launch Spectre both from OCEAN and from ADE.  I use different simulation setups in the two cases.  But sometimes when running ADE immediately after OCEAN, the input.scs file is not updated and the simulation is run with the setup in the OCEAN file.  If I do Simulation>Netlist>Recreate it does regenerate the netlist though.

Any ideas on why the "Netlist and Run" button in ADE is not working?
Back to top
 
 
View Profile   IP Logged
boe
Community Fellow
*****
Offline



Posts: 615

Re: problem with netlist recreation
Reply #1 - Feb 18th, 2010, 5:28am
 
Hi alirezad,
ADE does incremental netlisting, i.e. a new netlist is only generated if something has changed. Apparently in your case the ADE does not notice the change of setup.
BOE
Back to top
 
« Last Edit: Feb 19th, 2010, 2:29am by boe »  
View Profile   IP Logged
alirezad
Junior Member
**
Offline



Posts: 11

Re: problem with netlist recreation
Reply #2 - Feb 18th, 2010, 9:21am
 
Yes, You are right.  That is what it seems to be doing.  When OCEAN overwrites the netlist, ADE doesn't sense it and so the next time I run ADE it doesn't update the file.  However do you know if there is any way that I can override this setting, other than recreating the netlist from the Simulation>Netlist>Recreate menu and then running the simulation?  

I thought that is what the green "Netlist and Run" button is supposed to do.
Back to top
 
 
View Profile   IP Logged
alirezad
Junior Member
**
Offline



Posts: 11

Re: problem with netlist recreation
Reply #3 - Feb 18th, 2010, 9:30am
 
Thanks for your comment boe!  
Just knowing that the problem is because of the "incremental netlisting" helped me find this solution in the ADE known problems and solutions:

PCR 368310: ADE ignores changes in simulation and model files during incremental netlist and run in the same session

Description: When you edit a model file during a session, the Analog Design Environment may not be aware that the file was changed, and subsequent simulations may continue to use parameters from the old file. The same thing can happen with other files on which the netlist depends (auxiliary files like Definition files, Stimulus files and so on.)

Solution: Always Check and Save (using the Check and Save button in the Composer window) after changing any auxiliary file. This will force a re-netlist. You can also recreate the netlist but that may not take into account any unsaved changes in the design.
Back to top
 
 
View Profile   IP Logged
Andrew Beckett
Senior Fellow
******
Offline

Life, don't talk to
me about Life...

Posts: 1742
Bracknell, UK
Re: problem with netlist recreation
Reply #4 - Feb 19th, 2010, 8:02am
 
If you're only changing included files (and not editing them via the
ADE GUI), then that's not really anything to do with the fact that the
netlister is incremental. There are two parts to the simulator input
creation:

1. The netlist itself (which is netlisted from the design hierarchy)
2. The control statements etc which are added around this (inclusion of model files, parameters, analyses, outputs to save and so on).

Both of these should be triggered by a change in the source data for these parts (there have been bugs in the past, but I'm not aware of any outstanding problems).

The second issue is that spectre itself is normally run (in ADE or OCEAN) in what is called "interactive" mode - this means that the simulator does not exit at the end, and if a change in a design variable is made, it can do so in memory and quickly re-simulate without needing the simulator to start up, check out licenses, and parse the netlist. Bigger changes (resulting in a new netlist, or a new analysis, say) ADE will tell spectre to exit and re-read the new input.scs. If you just edit (in UNIX) one of the model files (say), spectre or ADE have no idea that it has changed, and consequently the simulation will be re-done with the same netlist.

Actually it's not necessary in that scenario to do a check-and-save, or a force renetlist. Just do Simulation->Stop and then the next time around spectre will definitely start again.

Your problem may be something else though - you didn't say what version you're using (type getVersion(t) in the CIW. or do Help->About) - maybe you have an older version that has a bug - maybe you should contact Cadence Customer Support if you're using something recent and this is still happening?

Regards,

Andrew.
Back to top
 
 
View Profile WWW   IP Logged
alirezad
Junior Member
**
Offline



Posts: 11

Re: problem with netlist recreation
Reply #5 - Feb 19th, 2010, 11:10am
 
I am using sub-version 5.10.41.500.5.105.

It makes sense that if I edit the simulator netlist myself, e.g. in unix, then ADE would know nothing about the change and the netlist will not be updated in the next run.  Now, if I have different analysis setups in OCEAN and in ADE then the netlist changes when I run OCEAN.  Next if I run ADE it still doesn't know about the change and doesn't recreate the netlist and just runs with OCEAN setup.

Actually for me if I stop the simulation and then run again it doesn't recreate the netlist.  Doing "check and save" does work anyways and it is easier than running, stopping and running again.

BTW, thanks Andrew for the comment about interactive mode.  Now I understand why I don't need to renetlist in parametric analysis, and why an idle ADE doesn't release the license... I feel like these are important things to know.  I wonder if there is a readable document that has information on "incremental netlisting" or "interactive mode" that you would recommend.
Back to top
 
 
View Profile   IP Logged
Andrew Beckett
Senior Fellow
******
Offline

Life, don't talk to
me about Life...

Posts: 1742
Bracknell, UK
Re: problem with netlist recreation
Reply #6 - Feb 20th, 2010, 8:26am
 
Do you have an open ADE session at the same time as running OCEAN? That would definitely confuse things, because they both thing they are in control...

Regards,

Andrew.
Back to top
 
 
View Profile WWW   IP Logged
alirezad
Junior Member
**
Offline



Posts: 11

Re: problem with netlist recreation
Reply #7 - Mar 2nd, 2010, 2:49pm
 
That's right.  I was transitioning from ADE to OCEAN so I had both open.
Thanks.
Back to top
 
 
View Profile   IP Logged
neoflash
Community Fellow
*****
Offline

Mixed-Signal
Designer

Posts: 397

Re: problem with netlist recreation
Reply #8 - Jan 19th, 2016, 11:42am
 
Hi Andrew,

Is there any way to force Spectre ADE stop doing "incremental netlisting" and always re-generate a new netlist to avoid such problem?

I have been confused by this issue for a while because some updates are not sensed by Cadence and gives me old results.

Thanks,
Neo

Andrew Beckett wrote on Feb 19th, 2010, 8:02am:
If you're only changing included files (and not editing them via the
ADE GUI), then that's not really anything to do with the fact that the
netlister is incremental. There are two parts to the simulator input
creation:

1. The netlist itself (which is netlisted from the design hierarchy)
2. The control statements etc which are added around this (inclusion of model files, parameters, analyses, outputs to save and so on).

Both of these should be triggered by a change in the source data for these parts (there have been bugs in the past, but I'm not aware of any outstanding problems).

The second issue is that spectre itself is normally run (in ADE or OCEAN) in what is called "interactive" mode - this means that the simulator does not exit at the end, and if a change in a design variable is made, it can do so in memory and quickly re-simulate without needing the simulator to start up, check out licenses, and parse the netlist. Bigger changes (resulting in a new netlist, or a new analysis, say) ADE will tell spectre to exit and re-read the new input.scs. If you just edit (in UNIX) one of the model files (say), spectre or ADE have no idea that it has changed, and consequently the simulation will be re-done with the same netlist.

Actually it's not necessary in that scenario to do a check-and-save, or a force renetlist. Just do Simulation->Stop and then the next time around spectre will definitely start again.

Your problem may be something else though - you didn't say what version you're using (type getVersion(t) in the CIW. or do Help->About) - maybe you have an older version that has a bug - maybe you should contact Cadence Customer Support if you're using something recent and this is still happening?

Regards,

Andrew.

Back to top
 
 
View Profile   IP Logged
Frank Wiedmann
Community Fellow
*****
Offline



Posts: 678
Munich, Germany
Re: problem with netlist recreation
Reply #9 - Jan 20th, 2016, 12:30am
 
See http://support.cadence.com/wps/mypoc/cos?uri=deeplinkmin:ViewSolution;DocumentTy....

I'm afraid that Andrew is no longer watching this forum (he has not posted here for a long time). In order to reach him, you should go to http://community.cadence.com/cadence_technology_forums/f/38.
Back to top
 
 
View Profile WWW   IP Logged
neoflash
Community Fellow
*****
Offline

Mixed-Signal
Designer

Posts: 397

Re: problem with netlist recreation
Reply #10 - Jan 20th, 2016, 10:58am
 
This works. Thanks a lot.

Frank Wiedmann wrote on Jan 20th, 2016, 12:30am:
See http://support.cadence.com/wps/mypoc/cos?uri=deeplinkmin:ViewSolution;DocumentTy....

I'm afraid that Andrew is no longer watching this forum (he has not posted here for a long time). In order to reach him, you should go to http://community.cadence.com/cadence_technology_forums/f/38.

Back to top
 
 
View Profile   IP Logged
Pages: 1
Send Topic Print
Copyright 2002-2024 Designer’s Guide Consulting, Inc. Designer’s Guide® is a registered trademark of Designer’s Guide Consulting, Inc. All rights reserved. Send comments or questions to editor@designers-guide.org. Consider submitting a paper or model.