Im thinking of using MDL for comparing waves coming from Verilog to waves coming from spectre.
Let's say I have a fsdb wave from verilog AMS simulation and I'd like to apply to it a set of measurements such as max, cross etc...
and compare results to measurements I obtained from spectremdl run. So my question is:
Is it possible to use mdl config file for postprocessing FSDB wave without netlist? Smth like:
mdl -b cfg.mdl -r my_design.fsdb -out my_meas_dir
All examples in MDL_workshop are about specte/ultrasim postprocessing when netlist is available. In my case, I just have a file with wave that should be self-sufficient.
You won't be able to process an FSDB database regardless, because it's not a Cadence format. So the tools can write it, but cannot read it.
If the output was psf/psfxl (say) then it's possible. I don't believe the netlist is required - it's only needed if your MDL references netlist parameters in the MDL code (as far as I know from a quick experiment)
In reply to Andrew Beckett:
so as far Im aware, cadence provides a way to convert vcd to psf and Springsoft has a utility that converts (digital only) fsdb to vcd.
Seems like a deadend for analog fsdb waves. I'll try to find a way to generate VCD waves though. If what you say is correct i.e. no netlist is needed and I can get all this rich MDL API to measure converted from VCD waves, it could be a workaround.