Help needed: Extract with NCSU_CDK.1.4 in Linux Fedora 3

Hi,

I installed NCSU_CDK 1.4 in Fedora 3.0. I use IC 5.0.33. When I do
extract on my Layout, I see the following error message. Does this
mean I didn't install NCSU library correctly? Can NCSU 1.4 work with
Fedora 3.0 and IC 5.0.33?

thanks,
Liang-Kai
================
\o library: cla64
\o cell: cla64
\o view: layout
\o Rules come from library NCSU_TechLib_ami06.
\o Rules path is divaEXT.rul.
\o Inclusion limit is set to 1000.
\o Switches used: Extract_parasitic_caps.
\o Parsing drcExtractRules of
"/crystal/eda/cadence/local/lib/NCSU_TechLib_ami06
/divaEXT.rul"...
\o error: Invalid input for multiLevelParasitic(); a cap list is
expected.
\o error: Illegal input layer capacitance found in saveParasitic().
\o error: The number of entries of property list and layer list must
be the sam
e.
\o saveParasitic(capacitance "PLUS" "MINUS" "c" "cap ivpcell
NCSU_Analog_Parts")
\o error: Invalid input for multiLevelParasitic(); a cap list is
expected.
\o error: Illegal input layer capacitance found in saveParasitic().
\o error: The number of entries of property list and layer list must
be the sam
e.
\o saveParasitic(capacitance "PLUS" "MINUS" "c" "cap ivpcell
NCSU_Analog_Parts")
\o error: Invalid input for multiLevelParasitic(); a cap list is
expected.
\o error: Illegal input layer capacitance found in saveParasitic().
\o error: The number of entries of property list and layer list must
be the sam
e.
\o saveParasitic(capacitance "PLUS" "MINUS" "c" "pcapacitor ivpcell
NCSU_Analog_
Parts")
\o error: Invalid input for multiLevelParasitic(); a cap list is
expected.
\o error: Illegal input layer capacitance found in saveParasitic().
\o error: The number of entries of property list and layer list must
be the sam
e.
\o saveParasitic(capacitance "PLUS" "MINUS" "c" "pcapacitor ivpcell
NCSU_Analog_
Parts")
\o info: If short location check is desired later on, check the
saveInterconnect
\o statement for correctness:
\o Interconnect layer nwell, pwell, pBulk not saved.
\o Interconnect layers merged on save: elecHighres, elecRes, polyRes,
nwellRes.
\o Interconnect layers merged on save: cp, pDiffContact,
nDiffContact, pOhmicCo
ntact, nOhmicContact, ce.
\o Interconnect layers merged on save: pDiff, nDiff, pOhmic, nOhmic.
\o
\o Errors exist in the rules file
"/crystal/eda/cadence/local/lib/NCSU_TechLib_a
mi06/divaEXT.rul".
\o Verification program terminated.
\r t
\r t

 

On Mon, 24 Oct 2005 03:10:09 -0500,

I took a look at the divaMultiLevel.il you posted to chiptalk and was
able to reproduce the error mentioned in the discussion, by forgetting
to run all the layers through geomConnect. Did you have something after
the "Errors exist in the rules file" line? It might direct you to the
source of the failure.

On a side note, I noticed that there are two cap clauses for the same
layer pair, using the same parameters. This will double the capacitance.

Parsing drcExtractRules of
"/usr2/ekalenda/pcrdata/example/Lib/divaEXT.rul"...
error: Illegal input layer capacitance found in saveParasitic().
error: The number of entries of property list and layer list must be
the same.
saveParasitic(capacitance "PLUS" "MINUS" "c" "pcapacitor symbol
analogLib")
Errors exist in the rules file
"/usr2/ekalenda/pcrdata/example/Lib/divaEXT.rul".
*WARNING* multiLevelParasitic() - This cannot be used before the
'geomConnect' command.
Verification program terminated.


Ed "Mr. Diva" Kalenda

 

Thanks for the test case. I was able to find the problem and adjust the
rules compiler to handle it.

Every coefficient is zero for elec to metal2. The rule compiler was
changed to skip the processing steps with a coefficient of zero since
they won't produce anything and was skipping them all. This caused the
message about an invalid multiLevelParasitic rule.

The techfile for the PDK should be corrected to have meaningful values,
but that's up to the NCSU people.

The rule deck is also odd in that it has duplicate cap clauses for elec
to metal2, and for elec to metal1.

Ed "Mr. Diva" Kalenda
Cadence Design Systems

This is just me blathering, not the company, since they don't let talk for them.

 

Hi Ed,

thanks so much for fixing this Diva problem.

I posted a patch to the NCSU PDK 1.4 here:
http://www.chiptalk.org/modules/newbb/viewtopic.php?topic_id=53

To summarize, Ed fixed 2 bugs in the NCSU PDK:
- Syntax error in divaDRC.rul
- Diva Extraction fails under IC51-41

Patches for both bugs are provided at the above website.

Thanks again,
Johannes Grad