When I am using APS for PSS (shooting Method) using swap file, APS is being disbled.
How can prevent APS being disabled.
Are you invoking APS in your ocean script? Specifically, have you included a line such as:
envOption( 'userCmdLineOption "+lqt 0 +lqtimeout 0 -64 +aps")
This line invokes aps (+aps) as well as uses a 64 bit binary and waits indefinitely for a spectre license to be available. I have not experienced an issue invoking aps in a PSS simulation.
In reply to smlogan:
No, I am NOT invoiking through ocean script. It ia ADE-RUN.
In reply to RFStuff :
You can invoke aps from the ADE. Under the Setup->Environment panel, add "+aps" to userCmdLineOption as shown below:
I am not having problem invoking APS.
The issue is : APS is some how getting disabled by Spectre on-the-fly for PSS-Shootinmg method with SWAP file.
The answer is simple; don't use swapfile. Swap file was an option added a long time ago to increase the capacity of SpectreRF when memory was a lot more expensive than it is now, and there was no 64 bit version of spectre. Nowadays you'd be far better off using 64 bit spectre if you need the capacity.
It always had a negative effect on performance because it had to write all the information calculated during the shooting interval to disk - which could be a lot of data. In the past if you really needed to use this you should only use it if absolutely necessary and then make sure that the swapfile was on a local (not networked) disk.
Now, turn on 64 bit either via the option on the setup->environment form or by typing -64 in the userCmdLineOption field on that form.
In reply to Andrew Beckett:
I did without using SWAP file and APS is enabled.
By the way, I am getting following notice:-
Notice from spectre during IC analysis, during periodic steady state analysis `pss'. Bad pivoting is found during DC analysis. Option dc_pivot_check=yes is recommended for possible improvement of convergence.
This is certainly not due to APS.
But I want to know why it is happening and where I can find "Option dc_pivot_check=yes" to make it "yes". What will be its effect.
I have attached the spectre.out file.
Could you please have a look at it tell me why the notice is cropping up.
Unfortunately this option is not well documented yet - there's a CCR to get that fixed. It's described in "spectre -h options" as:
112 dc_pivot_check=no During DC analysis numeric pivoting is only performed when bad pivot is detected. Possible values are no and yes.
You can enter this by going to Simulation->Options->Analog and then at the bottom of the form (or the Misc tab in IC61X) in the Additional Parameters field, enter "dc_pivot_check=yes".
I'm not sure it's that important though... it might speed up the convergence during the DC analysis though.
Thanks a lot for your reply.
Now I am running into a totally different problem called PSS convergence issue. I am thinking to file a SR but they process SR ( given the past experience) very slowly and it may take few weeks.
If you can help it will be really great as running far behind the scheduled timeline.
The problem is:-
With input port (dc) the PSS is converging.
But when I am adding two large signal tones, the PSS is not able to coverge.
Could you please help what I should do make it coverge. If you want any files, I can post/send.
Please contact customer support - I really don't have the bandwidth to debug every user's convergence problems in these forums (I am answering here in my (quite limited) spare time). If you stress the urgency, it can be prioritized. It really should not take that long...
Yes, it is really time consuming to debug convergence problem and I know that you have limited spare time. However, I am really thankful to you for promptly answering my queries in this forum inspite of your limited spre time.
I will be filing and SR and hoping that the ckt converges ASAP.