files missing in Spectre psf result directory when running simulation from command line

Hans,

Presumably you're not running it via the run() function in OCEAN, because that would then write out the variables database and set up the runObjFile to reference it. How are you running the simulation?

If you had an info analysis in the netlist:

designParamVals info what=parameters where=rawfile

Then you'd be able to access the variables using:

pv("top-level" "CAP" ?result 'designParamVals)

where CAP is the variable name. No need for the other stuff, which is additional info that ADE/OCEAN outputs.

Regards,

Andrew.

Andrew,

Thanks for quick response!

Indeed, the script submits one or more jobs using  ipcBeginProcess(run_string "")  instead of the Ocean Run().

And then ipcWait() before starting the post processing.

For the Spectre simulator the input.scs contains "designParamVals info what=parameters where=rawfile" allowing the expression pv("top-level" "CAP" ?result 'designParamVals). But I have one other simulator that does not generate that info...

Just to be sure: I went through Skill and Ocean language reference manuals in search of a function that generates the files missing in the psf. I've observed they are generated/written during run time. But did not find anything. The file creation is undocumented ?

kind regards

Hans

Hans,

There's no standalone function to do this, because it's intended to be done as part of ADE or OCEAN running the simulation. If you're bypassing that, you're also bypassing the component of ADE which records the design variables it has passed to the simulator - because it didn't know you'd even called the simulator.

Why don't you just use run()?

Regards,

Andrew.

Andrew,

My thinking was that even when "bypassing" the ADE framework  there could have been an oasis function that generates the files now missing in psf, given that e.g. netlist + psf directories and simulator are all known the moment the simulator is called ? I would just need to call that function in parallel and drop the files into the psf.

The ipcBeginProcess()  for running the simulator was inherited.... You are saying that run() can be a full equivalent, even when running third party simulator and interactively running multiple parallel simulations ?  

kind regards,

Hans

Andrew,

My thinking was that even when "bypassing" the ADE framework  there could have been an oasis function that generates the files now missing in psf, given that e.g. netlist + psf directories and simulator are all known the moment the simulator is called ? I would just need to call that function in parallel and drop the files into the psf.

The ipcBeginProcess()  for running the simulator was inherited.... You are saying that run() can be a full equivalent, even when running third party simulator and interactively running multiple parallel simulations ?  

kind regards,

Hans

Hans,

Not sure why there would need to be a public function to do this. It's part of ADE and the expectation is that there would be no need to run it separately. So there isn't a public function to do this as a separate step.

If you can run the third party simulator in ADE, you can run it using run(). Parallel simulations are possible via the distributed mechanism (you can use the command-line mode if you're not really using a distributed mechanism). The variables database is a bit of a legacy thing anyway.

Regards,

Andrew.