Affirma: How to save spectre.fc in parametric simulation

[Svetozar Broussev]

Hello,

I would like to save the transient final conditions (spectre.fc by default)
of a circuit after running a parametric simulation. I would like to have the
results for each run separately, so I can run after that pss and pnoise
analysis whenever I want to.

What Affrima does, it just saves one single file spectre.fc, perhaps the
last run. Is there any simple way to make the parametric simulation and have
at the end different *.fc files?

Thanks,
Svetozar
--

 

[Andrew Beckett]

Hi Svetozar,

You can use some of the magic characters that spectre supports in file
name specifications.

In general with spectre native sweeps this works well - I would tend
to specify my writefinal as "%A.fc" - where %A is the analysis name.
However, if you're using the parametric analysis tool in ADE this
doesn't work so well, since all points in the sweep name the same
name.

You could in theory use %T (time) or %P (process id) but these are
the same for the whole simulation, since by default spectre does not
restart for each point in a parametric run.

So if you really want to do this, you can do it by doing:

envSetVal("spectre.envOpts" "controlMode" 'string "batch")

this needs to be done before you start ADE. Then in the writefinal
specification, set it to (say)

%A_%T_%P.fc

Regards,

Andrew.

On Thu, 7 Oct 2004 17:54:37 +0300, "Svetozar Broussev"
<svetozar@spam.no> wrote:

Hello,

I would like to save the transient final conditions (spectre.fc by default)
of a circuit after running a parametric simulation. I would like to have the
results for each run separately, so I can run after that pss and pnoise
analysis whenever I want to.

What Affrima does, it just saves one single file spectre.fc, perhaps the
last run. Is there any simple way to make the parametric simulation and have
at the end different *.fc files?

Thanks,
Svetozar

 

[Andrew Beckett]

On Tue, 19 Oct 2004 17:50:35 +0300, "Svetozar Broussev"
<svetozar@spam.no> wrote:

The bad thing is that each run overwrites the data
from the previous run, probably this could also be solved somehow.

You can change the results directory by using resultsDir(), to stop it
overwriting the data from previous runs.

Andrew.

 

[Svetozar Broussev]

Thanks Andrew,

I finally get it working. Running spectre in batch mode gives different *.fc
files. The bad thing is that you have to know in advance what you are
planning to do, before you start ADE.

Meanwhile I found another solution, using ocean script. Although I haven't
been ocean before, I easily created my own script. Using the Result browser,
I located the original ocean script generated by ADE, and just by copy/paste
I created many single runs having different options, including different
*.fc files. It also gives a great freedom in running simulations in
non-parametric fashion. The bad thing is that each run overwrites the data
from the previous run, probably this could also be solved somehow.
Anyway, I got what I wanted to.

Regards
Svetozar

"Andrew Beckett" <andrewb@DcEaLdEeTnEcTe.HcIoSm> wrote in message
news:tkgqm0143p5uall0ps8c88e41mh4ofhbj5@4ax.com...

Hi Svetozar,

You can use some of the magic characters that spectre supports in file
name specifications.

In general with spectre native sweeps this works well - I would tend
to specify my writefinal as "%A.fc" - where %A is the analysis name.
However, if you're using the parametric analysis tool in ADE this
doesn't work so well, since all points in the sweep name the same
name.

You could in theory use %T (time) or %P (process id) but these are
the same for the whole simulation, since by default spectre does not
restart for each point in a parametric run.

So if you really want to do this, you can do it by doing:

envSetVal("spectre.envOpts" "controlMode" 'string "batch")

this needs to be done before you start ADE. Then in the writefinal
specification, set it to (say)

%A_%T_%P.fc

Regards,

Andrew.

On Thu, 7 Oct 2004 17:54:37 +0300, "Svetozar Broussev"
svetozar@spam.no> wrote:

Hello,

I would like to save the transient final conditions (spectre.fc by
default)
of a circuit after running a parametric simulation. I would like to have
the
results for each run separately, so I can run after that pss and pnoise
analysis whenever I want to.

What Affrima does, it just saves one single file spectre.fc, perhaps the
last run. Is there any simple way to make the parametric simulation and
have
at the end different *.fc files?

Thanks,
Svetozar