basem
New Member
Offline
Posts: 5
The sun valley.
|
Hello,
I have a mixed-signal circuit. I use pss successfully (Cadence) and attain a periodic steady state.
The PSS run uses multiple threading and about 10GB+ of memory.
When the pnoise start (which is set for wide frequency range), the simulation time is at least one order of magnitude longer (in fact, after a very long wait, the sim is still running at 80% completion). I checked and found out that pnoise uses single processor to run the circuit whereas pss uses 8 of them and lots of memory.
I think I am using the latest of the old Cadence software (IC5141) and I think I am running mmsim with APS option in the ADE activated.
The sim is running the 64 bit binary.
I am puzzled on why Cadence settled with non-parallel processing for pnoise?! It is not always the case that we need period noise analysis at a narrow band of operation!!
I have to now compromise on what I want to measure in the simulator to have a reasonable overall simulation time when I need to run a large number of these sims across corners and mismatch.
Any ideas? experiences? feedback? is highly appreciated.
And I can't use some other tool. I am stuck with this tool for this circuit.
Thank you :)
Basem
|