A stochastic model in Verilog A for Monte Carlo simulation

The first problem is that your "source.scs" should not have the ahdl_include statement in it. Since you've created a cellView in Virtuoso for that cell containing the veriloga code, the spectre netlister will automatically add the "ahdl_include" for you. So you've defined it twice because of this.

The second issue is that (for some reason) the instance of source in your schematic seems to be passing seedin - which it shouldn't be. Did you set the value on the instance? I suspect that maybe there was an issue with the veriloga parser in MMSIM10.1 which meant that it mistakenly identified this cds_inherited_param as a normal parameter - I don't have that version to hand though so I can't test that.

Could you use Tools->CDF->Edit in the CIW, set it to "Base", and pick the lib and "source" cell that you've created, and then use the "CDF Dump" button to dump out the CDF for the "source" cell, and then post it here (maybe as an attachment via the Options tab)? Then I can see what's going on in your case.

Kind Regards,

Andrew.

 Hello Andrew,

I have modified the file "source.scs" as following:

simulator lang=spectre
parameters seedin=0
statistics {
  mismatch {
    vary seedin dist=gauss std=5
  }
}

And the file "veriloga.va" is:

`include "constants.vams"
`include "disciplines.vams"
module source(a,b);
     inout a,b;
     electrical a,b;
     integer seed;
     (*cds_inherited_parameter*)parameter real seedin = 0;
     analog begin
     seed=seedin;
     V(b) <+ cos($abstime*seed*100000000);
     end
endmodule

When I did as you said for "CDF", there came the error attached.

Best regards,

UU

From the error messages, you appear to have dumped the CDF and then added it as a Model Library in ADE. I have no idea why you did that, since I didn't ask you to (and clearly spectre would not be able to read a SKILL dump of the CDF - that makes no sense).

I asked you to dump the CDF and then post it in this forum so that I can see it to understand what is wrong...

Regards,

Andrew.

 Sorry, perhaps it's this one

You appear to have overwritten the "source.scs" file with your CDF dump! Not sure why you did that - dump it to any file name you like (eg "dump.txt" and then post the contents here. That's what I asked for, not a picture of the CDF form. 

You'll also need to recreate your source.scs file since you clobbered it.

Andrew 

Hello, 

I have done as you said, but there is nothing in the file "dump.txt" in the end. 

 Hello,

I'am sorry I have made an error. The text in the "dump.txt" is:

 
/****************************************************/
 LIBRARY = "PMAMTJ40symstoRp"
 CELL    = "source"
/****************************************************/

let( ( libId cellId cdfId )
    unless( cellId = ddGetObj( LIBRARY CELL )
        error( "Could not get cell %s." CELL )
    )
    when( cdfId = cdfGetBaseCellCDF( cellId )
        cdfDeleteCDF( cdfId )
    )
    cdfId  = cdfCreateBaseCellCDF( cellId )

    ;;; Simulator Information
    cdfId->simInfo = list( nil )
    cdfId->simInfo->ams = '( nil
        isPrimitive       ""
        extraTerminals    ""
        propMapping       ""
        termMapping       ""
        termOrder         ""
        componentName     ""
        excludeParameters ""
        arrayParameters   ""
        stringParameters  ""
        referenceParameters ""
        enumParameters    ""
        instParameters    ""
        otherParameters   ""
        netlistProcedure  ""
    )
    cdfId->simInfo->auCdl = '( nil
        dollarEqualParams ""
        dollarParams      ""
        modelName         ""
        namePrefix        ""
        propMapping       ""
        termOrder         ""
        componentName     ""
        instParameters    ""
        otherParameters   ""
        netlistProcedure  ""
    )
    cdfId->simInfo->auLvs = '( nil
        namePrefix        ""
        permuteRule       ""
        propMapping       ""
        deviceTerminals   ""
        termOrder         ""
        componentName     ""
        instParameters    ""
        otherParameters   ""
        netlistProcedure  ""
    )
    cdfId->simInfo->hspiceD = '( nil
        opParamExprList   ""
        optParamExprList  ""
        propMapping       ""
        termMapping       ""
        termOrder         ""
        namePrefix        ""
        componentName     ""
        instParameters    ""
        otherParameters   ""
        netlistProcedure  ""
    )
    cdfId->simInfo->spectre = '( nil
        modelParamExprList ""
        optParamExprList  ""
        opParamExprList   ""
        stringParameters  ""
        propMapping       ""
        termMapping       ""
        termOrder         ""
        componentName     ""
        instParameters    ""
        otherParameters   ""
        netlistProcedure  ""
    )

    ;;; Properties
    cdfId->formInitProc            = ""
    cdfId->doneProc                = ""
    cdfId->buttonFieldWidth        = 340
    cdfId->fieldHeight             = 35
    cdfId->fieldWidth              = 350
    cdfId->promptWidth             = 175
    cdfId->instNameType            = "schematic"
    cdfId->instDisplayMode         = "cellName"
    cdfId->netNameType             = "schematic"
    cdfId->termSimType             = "DC"
    cdfId->termDisplayMode         = "none"
    cdfId->paramSimType            = "DC"
    cdfId->paramDisplayMode        = "parameter"
    cdfId->viewInfo                = '(nil veriloga (nil termMapping (nil c "\\:c" b "\\:b"
a "\\:a"
) moduleName "source"
namePrefix "ahdl" termOrder ("a" "b") parameterList
(seedin)
)
)
    cdfSaveCDF( cdfId )
)

Thanks. I assume you've already removed the "seedin" CDF parameter (since it's not in the dump above), so it may well work already, but if you still have a problem, type this in the CIW:

almSetParameterList("PMAMTJ40symstoRp" "source" nil ?view "veriloga")

This will remove the "seedin" parameter from being potentially netlisted when the veriloga view is used. Note that you will still need to fix your source.scs file if it still contains a copy of the above dump information and replace it with the "statistics" block definition etc.

I'm pretty certain that the problem was due to the spectre version you're using causing the parameter with the cds_inherited_parameter attribute being identified as a normal parameter when the veriloga was parsed. Ah yes, it was - CCR 922010 was filed (by me) for this issue, and it was fixed in MMSIM11.1 ISR17. No need to switch versions of spectre if you use the workaround above - see this post.

Regards,

Andrew.

 Hello Andrew,

Thank you for your help and patience. In fact it didn't work, but now it works well with your command:

almSetParameterList("PMAMTJ40symstoRp" "source" nil ?view "veriloga")

I don't really understand why although I have read the lien that you sent me but I will try to learn more. By the way what does CRR means?

Best regards,

WANG You

A CCR is a Cadence Change Request and is the system we use for tracking bugs and enhancement requests. 

If you'd used a newer simulator version when editing the veriloga view, the problem would not have happened.

Regards,

Andrew