ocean gives different results for dfw_v4.4.3 and ic_v5.0.32.

S

Steven Dupont

Guest
Hi there,

I have designed a standard cell library and characterized it using ocean
dfw_v4.4.3 on HP-UX B.11.00.
Characterising it again, using exactly the same scripts etc. and
platform, but now using ic_v5.0.32 gives characterisation results that
deviate up to 5 % of those obtained with dfw_v4.4.3.

Can somebody explain??

Steven


--

Met vriendelijke groet

Steven Dupont


*********************************************
IMEC
Kapeldreef 75
B-3001 Leuven, Belgium

Ir. Steven Dupont
Project Engineer ASIC Design
Division INVOMEC

Tel.: +32 16 281932
Fax : +32 16 281584

mailto:Steven.Dupont@imec.be
http://www.imec.be/
**********************************************

This e-mail and/or its attachments may contain confidential
information. It is intended solely for the intended addressee(s).
Any use of the information contained herein by other persons is
prohibited. IMEC vzw does not accept any liability for the contents of
this mail and/or its attachments.
 
Steven,

It could be that the models you're using (such as bsim3v3?) have changed their
default version between the two versions - if you don't have a specific version
set in the model - actually, I just checked, and the default version
for bsim3v3 is still 3.1, so that's not it.. Also, sometimes there may have been
model and simulator bugs
that have been fixed (it depends what you're measuring).

So it's very hard to know without a specific example.

Regards,

Andrew.

On Thu, 15 Apr 2004 17:04:33 +0200, Steven Dupont <Steven.Dupont@imec.be> wrote:

Hi there,

I have designed a standard cell library and characterized it using ocean
dfw_v4.4.3 on HP-UX B.11.00.
Characterising it again, using exactly the same scripts etc. and
platform, but now using ic_v5.0.32 gives characterisation results that
deviate up to 5 % of those obtained with dfw_v4.4.3.

Can somebody explain??

Steven
Click to expand...
--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
 
Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for differences? Or
do I have to compare the manuals?

But even when defaults are all the same I assume it is still possible
that implementation of some parameters changed. So comparing defaults
and seeing no differences does not mean anything at all, I think.


Steven




Andrew Beckett wrote:
Steven,

It could be that the models you're using (such as bsim3v3?) have changed their
default version between the two versions - if you don't have a specific version
set in the model - actually, I just checked, and the default version
for bsim3v3 is still 3.1, so that's not it.. Also, sometimes there may have been
model and simulator bugs
that have been fixed (it depends what you're measuring).

So it's very hard to know without a specific example.

Regards,

Andrew.

On Thu, 15 Apr 2004 17:04:33 +0200, Steven Dupont <Steven.Dupont@imec.be> wrote:


Hi there,

I have designed a standard cell library and characterized it using ocean
dfw_v4.4.3 on HP-UX B.11.00.
Characterising it again, using exactly the same scripts etc. and
platform, but now using ic_v5.0.32 gives characterisation results that
deviate up to 5 % of those obtained with dfw_v4.4.3.

Can somebody explain??

Steven


--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
Click to expand...

--

Met vriendelijke groet

Steven Dupont


*********************************************
IMEC
Kapeldreef 75
B-3001 Leuven, Belgium

Ir. Steven Dupont
Project Engineer ASIC Design
Division INVOMEC

Tel.: +32 16 281932
Fax : +32 16 281584

mailto:Steven.Dupont@imec.be
http://www.imec.be/
**********************************************

This e-mail and/or its attachments may contain confidential
information. It is intended solely for the intended addressee(s).
Any use of the information contained herein by other persons is
prohibited. IMEC vzw does not accept any liability for the contents of
this mail and/or its attachments.
 
Steven Dupont wrote:

Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for differences? Or
do I have to compare the manuals?
Click to expand...
Set the following variable via the CIW:

envSetVal("asimenv.misc" "saveDefaultsToOCEAN" 'boolean t)

If you then create an OCEAN script in the ADE with "Session - Save
Script ...", all default values of the simulation are included.


-Thomas
 
I don't think this is what Steven meant.

One thing you could do is to create a file, dumpModels.scs
which contains the following line:

dumpModels info what=models where=file file=dumpModels.out

and then specify this as a model file or an include file.

Normally spectre will write out model parameter information into
a binary (psf) results file, but by adding the above, it will write it to
an ASCII file (in the netlist directory).

Not sure whether it will help - but it might...

Andrew.

On Tue, 20 Apr 2004 16:40:38 +0200, Thomas Popp <Thomas.Popp@iaik.tugraz.at>
wrote:

Steven Dupont wrote:

Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for differences? Or
do I have to compare the manuals?

Set the following variable via the CIW:

envSetVal("asimenv.misc" "saveDefaultsToOCEAN" 'boolean t)

If you then create an OCEAN script in the ADE with "Session - Save
Script ...", all default values of the simulation are included.


-Thomas
Click to expand...
--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
 
What I meant is : all parameters that are used by spectre and more in
detail those that affect accuracy of a simulation. Indeed as Andrew
suggested, model parameters belong to this group but e.g. also all
(default) parameters used by the tran statement or the option statement
: parametetrs as reltol, errpreset, iabstol etc.

Is it possible to have a 'dump'of these? The solution of Thomas only
lists the values you have entered explicetly yourself or the one that
artist sets itself. The simlator logfile seems to log the 'important' ones.

Steven

Andrew Beckett wrote:
I don't think this is what Steven meant.

One thing you could do is to create a file, dumpModels.scs
which contains the following line:

dumpModels info what=models where=file file=dumpModels.out

and then specify this as a model file or an include file.

Normally spectre will write out model parameter information into
a binary (psf) results file, but by adding the above, it will write it to
an ASCII file (in the netlist directory).

Not sure whether it will help - but it might...

Andrew.

On Tue, 20 Apr 2004 16:40:38 +0200, Thomas Popp <Thomas.Popp@iaik.tugraz.at
wrote:



Steven Dupont wrote:


Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for differences? Or
do I have to compare the manuals?

Set the following variable via the CIW:

envSetVal("asimenv.misc" "saveDefaultsToOCEAN" 'boolean t)

If you then create an OCEAN script in the ADE with "Session - Save
Script ...", all default values of the simulation are included.


-Thomas


--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
Click to expand...

--

Met vriendelijke groet

Steven Dupont


*********************************************
IMEC
Kapeldreef 75
B-3001 Leuven, Belgium

Ir. Steven Dupont
Project Engineer ASIC Design
Division INVOMEC

Tel.: +32 16 281932
Fax : +32 16 281584

mailto:Steven.Dupont@imec.be
http://www.imec.be/
**********************************************

This e-mail and/or its attachments may contain confidential
information. It is intended solely for the intended addressee(s).
Any use of the information contained herein by other persons is
prohibited. IMEC vzw does not accept any liability for the contents of
this mail and/or its attachments.
 
test

Steven Dupont wrote:
What I meant is : all parameters that are used by spectre and more in
detail those that affect accuracy of a simulation. Indeed as Andrew
suggested, model parameters belong to this group but e.g. also all
(default) parameters used by the tran statement or the option statement
: parametetrs as reltol, errpreset, iabstol etc.

Is it possible to have a 'dump'of these? The solution of Thomas only
lists the values you have entered explicetly yourself or the one that
artist sets itself. The simlator logfile seems to log the 'important'
ones.

Steven

Andrew Beckett wrote:

I don't think this is what Steven meant.

One thing you could do is to create a file, dumpModels.scs
which contains the following line:

dumpModels info what=models where=file file=dumpModels.out
and then specify this as a model file or an include file.

Normally spectre will write out model parameter information into
a binary (psf) results file, but by adding the above, it will write it to
an ASCII file (in the netlist directory).

Not sure whether it will help - but it might...

Andrew.

On Tue, 20 Apr 2004 16:40:38 +0200, Thomas Popp
Thomas.Popp@iaik.tugraz.at
wrote:



Steven Dupont wrote:


Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for differences? Or
do I have to compare the manuals?


Set the following variable via the CIW:

envSetVal("asimenv.misc" "saveDefaultsToOCEAN" 'boolean t)

If you then create an OCEAN script in the ADE with "Session - Save
Script ...", all default values of the simulation are included.


-Thomas



--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
Click to expand...

--

Met vriendelijke groet

Steven Dupont


*********************************************
IMEC
Kapeldreef 75
B-3001 Leuven, Belgium

Ir. Steven Dupont
Project Engineer ASIC Design
Division INVOMEC

Tel.: +32 16 281932
Fax : +32 16 281584

mailto:Steven.Dupont@imec.be
http://www.imec.be/
**********************************************

This e-mail and/or its attachments may contain confidential
information. It is intended solely for the intended addressee(s).
Any use of the information contained herein by other persons is
prohibited. IMEC vzw does not accept any liability for the contents of
this mail and/or its attachments.
 
test

Steven Dupont wrote:
What I meant is : all parameters that are used by spectre and more in
detail those that affect accuracy of a simulation. Indeed as Andrew
suggested, model parameters belong to this group but e.g. also all
(default) parameters used by the tran statement or the option statement
: parametetrs as reltol, errpreset, iabstol etc.

Is it possible to have a 'dump'of these? The solution of Thomas only
lists the values you have entered explicetly yourself or the one that
artist sets itself. The simlator logfile seems to log the 'important'
ones.

Steven

Andrew Beckett wrote:

I don't think this is what Steven meant.

One thing you could do is to create a file, dumpModels.scs
which contains the following line:

dumpModels info what=models where=file file=dumpModels.out
and then specify this as a model file or an include file.

Normally spectre will write out model parameter information into
a binary (psf) results file, but by adding the above, it will write it to
an ASCII file (in the netlist directory).

Not sure whether it will help - but it might...

Andrew.

On Tue, 20 Apr 2004 16:40:38 +0200, Thomas Popp
Thomas.Popp@iaik.tugraz.at
wrote:



Steven Dupont wrote:


Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for differences? Or
do I have to compare the manuals?


Set the following variable via the CIW:

envSetVal("asimenv.misc" "saveDefaultsToOCEAN" 'boolean t)

If you then create an OCEAN script in the ADE with "Session - Save
Script ...", all default values of the simulation are included.


-Thomas



--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
Click to expand...

--

Met vriendelijke groet

Steven Dupont


*********************************************
IMEC
Kapeldreef 75
B-3001 Leuven, Belgium

Ir. Steven Dupont
Project Engineer ASIC Design
Division INVOMEC

Tel.: +32 16 281932
Fax : +32 16 281584

mailto:Steven.Dupont@imec.be
http://www.imec.be/
**********************************************

This e-mail and/or its attachments may contain confidential
information. It is intended solely for the intended addressee(s).
Any use of the information contained herein by other persons is
prohibited. IMEC vzw does not accept any liability for the contents of
this mail and/or its attachments.
 
Steven Dupont wrote:
test

Steven Dupont wrote:

What I meant is : all parameters that are used by spectre and more in
detail those that affect accuracy of a simulation. Indeed as Andrew
suggested, model parameters belong to this group but e.g. also all
(default) parameters used by the tran statement or the option
statement : parametetrs as reltol, errpreset, iabstol etc.

Is it possible to have a 'dump'of these? The solution of Thomas only
lists the values you have entered explicetly yourself or the one that
artist sets itself. The simlator logfile seems to log the 'important'
ones.

Steven

Andrew Beckett wrote:

I don't think this is what Steven meant.

One thing you could do is to create a file, dumpModels.scs
which contains the following line:

dumpModels info what=models where=file file=dumpModels.out
and then specify this as a model file or an include file.

Normally spectre will write out model parameter information into
a binary (psf) results file, but by adding the above, it will write
it to
an ASCII file (in the netlist directory).

Not sure whether it will help - but it might...

Andrew.

On Tue, 20 Apr 2004 16:40:38 +0200, Thomas Popp
Thomas.Popp@iaik.tugraz.at
wrote:



Steven Dupont wrote:


Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether
there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for
differences? Or
do I have to compare the manuals?



Set the following variable via the CIW:

envSetVal("asimenv.misc" "saveDefaultsToOCEAN" 'boolean t)

If you then create an OCEAN script in the ADE with "Session - Save
Script ...", all default values of the simulation are included.


-Thomas




--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
Click to expand...
 
Steven,

No, I don't think you can dump the options and transient parameters (it does
output the important ones from a transient). That said, I don't think anything
has changed in that area.

I would suggest that you contact customer support about this - but you'll need
to provide your netlist/models for us to investigate (mention that we'd
discussed this, and I'll try to ensure that it is assigned to me - since you're
in Europe,
this should be no problem).

Regards,

Andrew.

On Thu, 22 Apr 2004 11:14:25 +0200, Steven Dupont <Steven.Dupont@imec.be> wrote:

What I meant is : all parameters that are used by spectre and more in
detail those that affect accuracy of a simulation. Indeed as Andrew
suggested, model parameters belong to this group but e.g. also all
(default) parameters used by the tran statement or the option statement
: parametetrs as reltol, errpreset, iabstol etc.

Is it possible to have a 'dump'of these? The solution of Thomas only
lists the values you have entered explicetly yourself or the one that
artist sets itself. The simlator logfile seems to log the 'important' ones.

Steven

Andrew Beckett wrote:
I don't think this is what Steven meant.

One thing you could do is to create a file, dumpModels.scs
which contains the following line:

dumpModels info what=models where=file file=dumpModels.out

and then specify this as a model file or an include file.

Normally spectre will write out model parameter information into
a binary (psf) results file, but by adding the above, it will write it to
an ASCII file (in the netlist directory).

Not sure whether it will help - but it might...

Andrew.

On Tue, 20 Apr 2004 16:40:38 +0200, Thomas Popp <Thomas.Popp@iaik.tugraz.at
wrote:



Steven Dupont wrote:


Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for differences? Or
do I have to compare the manuals?

Set the following variable via the CIW:

envSetVal("asimenv.misc" "saveDefaultsToOCEAN" 'boolean t)

If you then create an OCEAN script in the ADE with "Session - Save
Script ...", all default values of the simulation are included.


-Thomas


--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
Click to expand...
--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
 
Andrew,

Thanks for your info so far. The problem is that, as an europractice
member, I have no direct access to cadence customer support. I Have to
discuss this with my boss.

Steven

Andrew Beckett wrote:
Steven,

No, I don't think you can dump the options and transient parameters (it does
output the important ones from a transient). That said, I don't think anything
has changed in that area.

I would suggest that you contact customer support about this - but you'll need
to provide your netlist/models for us to investigate (mention that we'd
discussed this, and I'll try to ensure that it is assigned to me - since you're
in Europe,
this should be no problem).

Regards,

Andrew.

On Thu, 22 Apr 2004 11:14:25 +0200, Steven Dupont <Steven.Dupont@imec.be> wrote:


What I meant is : all parameters that are used by spectre and more in
detail those that affect accuracy of a simulation. Indeed as Andrew
suggested, model parameters belong to this group but e.g. also all
(default) parameters used by the tran statement or the option statement
: parametetrs as reltol, errpreset, iabstol etc.

Is it possible to have a 'dump'of these? The solution of Thomas only
lists the values you have entered explicetly yourself or the one that
artist sets itself. The simlator logfile seems to log the 'important' ones.

Steven

Andrew Beckett wrote:

I don't think this is what Steven meant.

One thing you could do is to create a file, dumpModels.scs
which contains the following line:

dumpModels info what=models where=file file=dumpModels.out

and then specify this as a model file or an include file.

Normally spectre will write out model parameter information into
a binary (psf) results file, but by adding the above, it will write it to
an ASCII file (in the netlist directory).

Not sure whether it will help - but it might...

Andrew.

On Tue, 20 Apr 2004 16:40:38 +0200, Thomas Popp <Thomas.Popp@iaik.tugraz.at
wrote:



Steven Dupont wrote:



Andrew, thanks for your reply.

The model version is explicitely stated to be 3.1 in the model itself,
so there is no discussion about that.

I also thought of checking all simulator defaults and see whether there
is any difference between defaults between dfw_v4.4.3 and ic_v5.0.32.
Do you know whether there is a means to 'dump' the simulator run
parameters, so that I can compare 'dumps' and look for differences? Or
do I have to compare the manuals?

Set the following variable via the CIW:

envSetVal("asimenv.misc" "saveDefaultsToOCEAN" 'boolean t)

If you then create an OCEAN script in the ADE with "Session - Save
Script ...", all default values of the simulation are included.


-Thomas


--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd


--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd
Click to expand...
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top