I expanded your uuencoded file, and it looks like it uses a Rfft star,
which is a contributed SDF star which does not have any demos.

It looks like the bug is a for loop in 
$PTOLEMY/src/domains/sdf/contrib/stars/SDFRfft.pl:

    Complex* Result = F.Compute( Temp1, Temp2 );

    double Norm = int(Inverse) ? WinNorm : WinNorm / (double(fftlength));
    for( i = 0; i < fftlength; i++ )
      Result[ i ] *= Norm;


I believe that the for loop should be:
    for( i = 0; i < fftlength/2; i++ )

This stops the crash, but I'm not sure if the results are correct.

Below is your pigilog.pt file
--start--
reset ___empty___
domain SDF
newuniverse fir2 SDF
target default-SDF
        targetparam logFile ""
        targetparam loopScheduler "DEF #choices: DEF, CLUST,ACYLOOP"
        targetparam schedulePeriod 0.0
        star IIDGaussian1 IIDGaussian
        setstate IIDGaussian1 mean 0.0
        setstate IIDGaussian1 variance 1.0
        pragma fir2 IIDGaussian1  
        star Rfft1 Rfft
        setstate Rfft1 InputSize 1024
        setstate Rfft1 fftSize 1024
        setstate Rfft1 Inverse 0
        setstate Rfft1 UseWindow 0
        pragma fir2 Rfft1  
        star FIR1 FIR
        setstate FIR1 taps "0.0075 0.024 0.0345 0.024 0.0075"
        setstate FIR1 decimation 1
        setstate FIR1 decimationPhase 0
        setstate FIR1 interpolation 1
        pragma fir2 FIR1  
        star CxToFloat1 CxToFloat
        pragma fir2 CxToFloat1  
        star BlackHole1 BlackHole
        pragma fir2 BlackHole1  
        connect IIDGaussian1 output FIR1 signalIn
        connect FIR1 signalOut Rfft1 input
        connect Rfft1 output CxToFloat1 input
        connect CxToFloat1 output BlackHole1 input
reset ___empty___
domain SDF
newuniverse fir2 SDF
target default-SDF
        targetparam logFile ""
        targetparam loopScheduler "DEF #choices: DEF, CLUST,ACYLOOP"
        targetparam schedulePeriod 0.0
        star IIDGaussian1 IIDGaussian
        setstate IIDGaussian1 mean 0.0
        setstate IIDGaussian1 variance 1.0
        pragma fir2 IIDGaussian1  
        star Rfft1 Rfft
        setstate Rfft1 InputSize 1024
        setstate Rfft1 fftSize 1024
        setstate Rfft1 Inverse 0
        setstate Rfft1 UseWindow 0
        pragma fir2 Rfft1  
        star FIR1 FIR
        setstate FIR1 taps "0.0075 0.024 0.0345 0.024 0.0075"
        setstate FIR1 decimation 1
        setstate FIR1 decimationPhase 0
        setstate FIR1 interpolation 1
        pragma fir2 FIR1  
        star CxToFloat1 CxToFloat
        pragma fir2 CxToFloat1  
        star BlackHole1 BlackHole
        pragma fir2 BlackHole1  
        connect IIDGaussian1 output FIR1 signalIn
        connect FIR1 signalOut Rfft1 input
        connect Rfft1 output CxToFloat1 input
        connect CxToFloat1 output BlackHole1 input
--end--

Thanks for reporting this, and thanks for supplying a small test case.

-Christopher
--------

    Hi,
    
    I've been looking at Ptolemy and I found that using the attached
    schematic consisting of pre-built models makes Ptolemy crash (RPC Exit)
    when running many iterations, like 1000, but that it works OK when
    running only 100.
    
    The same thing happens if I try to dump the signals in a file (using the
    model "Printer") instead of just killing the signal.
    
    My OS: HPUX10.20
    Compiler: egcs-1.0.2
    
    Best Regards,
    
    Henrik Balle
    
    begin 777 PIGILOG.PT
-deleted
---

----------------------------------------------------------------------------
Posted to the ptolemy-hackers mailing list.  Please send administrative
mail for this list to: [EMAIL PROTECTED]

Reply via email to