HELP: Not enough storage space when taking Transient simulation

[ccarrot]

Hi everyone:

I'm taking a long transient simulation in Cadence Spectre, the simulation is to calculate the BER(bit error rate) of a transistor-level circuit.



Because I have a symbol which is written in Verilog-AMS to calculate the BER and store the result in a text file outside, so I don't need the cadence to store any simulation result. I have done:

1. In ADE XL Test Editor, =>Output=>save all, I unclick all the selections (save nothing), and click OK

But at last, a file named "tran.tran" will be still saved in "~/cadence/simulation/***/adexl/result/data/Interactive.0/1/***/psf", and this file in my simulation will be as large as about 600Gbit (after only 1-day simulation, and the data is still not sufficient for me); the cadence then will say it "face a fatal error" when writing data to "tran.tran" file, there is not enough space to store the data.





I even deleted the "tran.tran" file when the simulation started, but the problem was still there.

Could anyone help me with these:

1. Why cadence still saves the transient data even thought I have selected "save nothing"? Did I make the wrong setup?

2. Is there any method can be taken to let cadence just do the simulation but save no simulation data?



I'm very appreciated for your help! I've trapped myself in this problem for days! Thank you very much for any help!


Thanks
ccarrot

 

[erikl]

Try the options outputstart=<end-time> or skipstart=0 , see View attachment Spectre_Circuit_Simulator_Reference_V5_p840-842.pdf .

If even this doesn't help, you could try -- after the tran.tran file has been created -- the Unix/Linux command
(from within the psf/ directory):

mv tran.tran /dev/null/

Good luck!

 

[ccarrot]

Try the options outputstart=<end-time> or skipstart=0 , see View attachment 76423 .

If even this doesn't help, you could try -- after the tran.tran file has been created -- the Unix/Linux command
(from within the psf/ directory):

mv tran.tran /dev/null/

Good luck!

Thank you so much erikl! It works! Thank you!

 

[erikl]

Thank you so much erikl! It works!

Hi ccarrot,

would you, possibly, reveal to us, which of the methods worked for you? This could help others!

Cheers, erikl

 

[ccarrot]

Hi ccarrot,

would you, possibly, reveal to us, which of the methods worked for you? This could help others!

Cheers, erikl

I used both methods for full guarantee. For example, I used the simulation time = 1s (run it for a month I don't care, as the results are recorded on a text file every half an hour), then I set:

outputstart=1 and skipstart=0

then the "tran.tran" file is always 0bits----nothing to be saved!

Thanks erikl again for his help!!

 

[erikl]

Thank you, ccarrot, for reporting back!