ocean script, sens analysis and ?addInstancesUsingWildcards option

Marcel,

Yes - we don't necessarily explicitly document all the options this way. However, if you use ocnHelp('analysis) you'll see it mentioned under the sens analysis. Also if you add the analysis, and do ocnDisplay('analysis) or ocnDisplay('analysis 'sens) you'll see the parameter listed there.

Or if you set it up in the ADE UI and export the OCEAN script, you'd see it there too.

Most of the time the arguments are named after the corresponding options in the simulator (as you would see from "spectre -h tran" for example), but in a few cases they are ADE-specific fields that need to be supported too.

Regards,

Andrew.

Marcel,

Yes - we don't necessarily explicitly document all the options this way. However, if you use ocnHelp('analysis) you'll see it mentioned under the sens analysis. Also if you add the analysis, and do ocnDisplay('analysis) or ocnDisplay('analysis 'sens) you'll see the parameter listed there.

Or if you set it up in the ADE UI and export the OCEAN script, you'd see it there too.

Most of the time the arguments are named after the corresponding options in the simulator (as you would see from "spectre -h tran" for example), but in a few cases they are ADE-specific fields that need to be supported too.

Regards,

Andrew.

Hi Andrew,

In ADE-L I've selected an instance (see image), and after that saved the state as ocean script.

But in the saved ocean script I have no list of selected instances:

analysis('sens ?anaName "sens"  ?analyses_list list("dc")  ?net list("/VDD_65V")  )
analysis('dc ?anaName "dc"  ?saveOppoint t  ?write "spectre.dc"
                ?oppoint "rawfile"  ?maxiters "50"  ?maxsteps "1000"  ?annotate "status"
                ?threshold "0.0"  ?detail "node"  ?sort "name"  )

Here is the CIW output of ocnDisplay('analysis 'sens): 

Analysis name: sens  Analysis type: sens
 (anaName "sens")
 (sensType (nil nil nil))
 (analyses_list ("dc"))
 outputs(nil)
 (output_list nil)
 (net ("/VDD_65V"))
 (term nil)
 (instances nil)
 (additionalParams "")
nil

getVersion(t)

"sub-version  IC6.1.7-64b.500.8.EHF2985 "

Do I need a newer version?

Marcel,

Two problems. Firstly the addInstancesUsingWildcards field was  only added in IC617 ISR9 (so just after what you're using). Secondly there was a bug which was fixed in IC617 ISR10 which meant that the state information wasn't being properly recorded (for the selected instances) - and so what that meant was that ADE XL didn't netlist the selected instances. I suspect (from knowing what was going on in the bug - I dealt with the issue myself) that it would also affect OCEAN too (I didn't check).

So I'd suggest trying this in an IC617 subversion greater than ISR10.

Regards,

Andrew.

Thank you Andrew.

The things looks better in the newer versions.

ocnDisplay('analysis 'sens)
Analysis name: sens  Analysis type: sens
 (anaName "sens")
 (sensType (nil nil nil))
 (analyses_list ("dc"))
 outputs(nil)
 (output_list nil)
 (net ("/vdd"))
 (term nil)
 (instances nil)
 (instancesList ("/M1" "/M0"))
 (addInstancesUsingWildcards "")
 (additionalParams "")

Best reagards,

Marcel

On last question on this thread.

I want to make the code compatible with older virtuoso version (e.g. "sub-version  IC6.1.6.500.13.2 ") without using wildcards.

What I intend to do is to get schematic insts names and to use a statement like below

  analysis('sens ?analyses_list list("dc")  ?net list("/out") ?instances hugeListOfInstNames)

Is this the right way?

I did some tests, the instance names were added to sens (...) to (...) for (dc) statement in input.scs file.

Also, is there a limit on the ocean script file size? just in case hugeListOfInstNames becomes a very long one.

Best Regards,

Marcel

Hi Marcel,

Yes, that sounds reasonable. Please double check that it works - because of this problem where the state information wasn't being updated properly (I think that specifying the ?instances should be OK though). Ah, just noticed that you said you'd specified this.

There's no limit on list size, nor in the size of an OCEAN script. There is a limit on the size of a SKILL form (e.g. the block of code for a let, a function definition) - but that's usually quite hard to reach in practice except in machine-generated code (and even then, I've not seen the limit being reached recently).

So it should be OK.

Regards,

Andrew.