spectre wrapper in ocean

S

stefano zanella

Guest
Hi,

I need to run spectre through a wrapper in ocean. In other words, I
need to do something like:


envSetVal("spectre.envOpts" "simExecName" 'string "mySpectreWrapper
spectre")
simulator('spectre)
.....
....
....

run()

By default spectre is run with +inter which starts MPS, and the
spectre process is started by my wrapper (is runSimulation started in
background?). The simulation never quits, I guess because it is waiting
for some MPS commands. The obvious workaround

envSetVal( "spectre.envOpts" "userCmdLineOption" 'string " -inter")

won't work either. The job runs and finishes correctly. However, ocean
then gives me the following error message:

Cannot start up spectre.
The directory you specified does not contain valid PSF
results. Please check the line for typographical errors or
look in the directory for a logFile and PSF result files.

I guess that this is somehow connected to the fact that runObjFile is
not created in the psf directory.

Any ideas?

Thanks,
Stefano
 
On Mon, 15 Aug 2005 12:21:16 -0700, stefano zanella <stefanoDOTzanella@pdf.com>
wrote:

Hi,

I need to run spectre through a wrapper in ocean. In other words, I
need to do something like:


envSetVal("spectre.envOpts" "simExecName" 'string "mySpectreWrapper
spectre")
simulator('spectre)
....
...
...

run()

By default spectre is run with +inter which starts MPS, and the
spectre process is started by my wrapper (is runSimulation started in
background?). The simulation never quits, I guess because it is waiting
for some MPS commands. The obvious workaround

envSetVal( "spectre.envOpts" "userCmdLineOption" 'string " -inter")

won't work either. The job runs and finishes correctly. However, ocean
then gives me the following error message:

Cannot start up spectre.
The directory you specified does not contain valid PSF
results. Please check the line for typographical errors or
look in the directory for a logFile and PSF result files.

I guess that this is somehow connected to the fact that runObjFile is
not created in the psf directory.

Any ideas?

Thanks,
Stefano
Rather than using that command line option, does it work if you use:

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

(the default is "interactive"). This makes spectre start a new run for every
simulation, rather than trying to continue in interactive mode.

The runObjFile is created by ADE after the simulation has completed, but it has
to know its finished... so the way you're doing it, ADE/OCEAN thinks it is
interactive, but the simulator thinks it isn't. So it's getting confused, I
expect.

Regards,

Andrew.
 
Hi Andrew,

Thanks a lot for your answer and explanation. I knew I was doing
something wrong! Now, stuff seems to be working (at least as I expect it
to do).

I guess I was supposed to RTFM. Any pointers in this direction?

Thanks a lot,
Stefano


Rather than using that command line option, does it work if you use:

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

(the default is "interactive"). This makes spectre start a new run for every
simulation, rather than trying to continue in interactive mode.

The runObjFile is created by ADE after the simulation has completed, but it has
to know its finished... so the way you're doing it, ADE/OCEAN thinks it is
interactive, but the simulator thinks it isn't. So it's getting confused, I
expect.

Regards,

Andrew.
 
Hi Stefano,

This is in:

Virtuoso Analog Design Environment User Guide --
Appendix A. Environment Variables

Andrew.

On Mon, 15 Aug 2005 17:03:18 -0700, stefano zanella <stefanoDOTzanella@pdf.com>
wrote:

Hi Andrew,

Thanks a lot for your answer and explanation. I knew I was doing
something wrong! Now, stuff seems to be working (at least as I expect it
to do).

I guess I was supposed to RTFM. Any pointers in this direction?

Thanks a lot,
Stefano



Rather than using that command line option, does it work if you use:

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

(the default is "interactive"). This makes spectre start a new run for every
simulation, rather than trying to continue in interactive mode.

The runObjFile is created by ADE after the simulation has completed, but it has
to know its finished... so the way you're doing it, ADE/OCEAN thinks it is
interactive, but the simulator thinks it isn't. So it's getting confused, I
expect.

Regards,

Andrew.
 
Hi Andrew,

I was looking for something more on the tutorial side. A reference is
good when you already know stuff, but it would be good to be able to
find a place where how these things work together.

Thanks,
Stefano


Andrew Beckett wrote:
Hi Stefano,

This is in:

Virtuoso Analog Design Environment User Guide --
Appendix A. Environment Variables

Andrew.

On Mon, 15 Aug 2005 17:03:18 -0700, stefano zanella <stefanoDOTzanella@pdf.com
wrote:


Hi Andrew,

Thanks a lot for your answer and explanation. I knew I was doing
something wrong! Now, stuff seems to be working (at least as I expect it
to do).

I guess I was supposed to RTFM. Any pointers in this direction?

Thanks a lot,
Stefano



Rather than using that command line option, does it work if you use:

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

(the default is "interactive"). This makes spectre start a new run for every
simulation, rather than trying to continue in interactive mode.

The runObjFile is created by ADE after the simulation has completed, but it has
to know its finished... so the way you're doing it, ADE/OCEAN thinks it is
interactive, but the simulator thinks it isn't. So it's getting confused, I
expect.

Regards,

Andrew.
 

Welcome to EDABoard.com

Sponsor

Back
Top