@User one minute to parse the megabytes of sky130 models is quite normal (of course it's system dependent). Ensure you have the following

.spiceinit

file in the directory where ngspice is running, the second line speeds up model loading.

set ngbehavior=hsa

set ng_nomodcheck

Hi @User, it's an honor. Thank you for your work. Ok, so I was thinking something like that... Let me check my

.spiceinit

...

When you run ngspice do you see this message in the simulator output? This ensures ngspice is reading the .spiceinit file. On my system ngspice start simulting after 30 seconds. My system is a 12 years old laptop, no SSD.

Compatibility modes selected: hs a

yes, I see that line:

Compatibility modes selected: hs a


Circuit: **.subckt miel21_tb_opamp_op

Scale set to 1e-06
Doing analysis at TEMP = 27.000000 and TNOM = 27.000000

 Reference value :  0.00000e+00
No. of Data Rows : 1

May be you should try to update ngspice from the sources https://git.code.sf.net/p/ngspice/ngspice

Here's the full netlist

on my system it takes 30 seconds, and 99% of the time is just parsing the huge pdk.

Right, I notice the same thing - the simulation is a snap, but the loading is taking all the time. Ok, thanks, so it's definitely something wrong with my setup, thanks.

yes one minute is bit too much, either there is a I/O bottleneck in the virtal machine or the ngspice version is too old. The load improvement is quite recent so let's hope things get better with updated ngspice.

ok, here we go:

******
** ngspice-36+ : Circuit level simulation program
** The U. C. Berkeley CAD Group
** Copyright 1985-1994, Regents of the University of California.
** Copyright 2001-2020, The ngspice team.
** Please get your ngspice manual from <http://ngspice.sourceforge.net/docs.html>
** Please file your bug-reports at <http://ngspice.sourceforge.net/bugrep.html>
** Creation Date: Fri May  6 09:49:12 UTC 2022
******

I also believe the ng_nomodcheck is now the default in ngspice. Even if i remove the .spiceinit file i don't see differences. However keep at least the compatibility option set ngbehavior=hsa.

ah, ok, so that's the thing I dropped out a few versions back and I just kept using the same line to compile, so the mistake propagated... Thank you

do not add --no-create --no-recursion, these are saved in my config.log for reasons i ignore, but they should not be given at config time.

Ok.

@User @User I made a script to put all the models into a single file, only keep a single corner, and strip all the comments from it. In my experience, this also brings quite a bit of speed. In case you are interested: https://github.com/hpretl/iic-osic

@User: Hello once again, and thanks for posting! To all of you over there in the European time zones: It's nice to find the complicated questions all cleared up before I've even had breakfast. To everyone: Aleksandar is a great guy and talented designer who has been active with efabless since the early days, and has always been very supportive of us, even when suffering through all of our start-up pains.

@User (@User) you are right, loading a single corner further speeds up start up time considerably. If instead of doing:

.lib /home/schippes/share/pdk/sky130A/libs.tech/ngspice/sky130.lib.spice tt

I take only the 'tt' portion of the .lib file: .

param mc_mm_switch=0

.param mc_pr_switch=0

* MOSFET

.include /home/schippes/share/pdk/sky130A/libs.tech/ngspice/corners/tt.spice

* Resistor/Capacitor

.include /home/schippes/share/pdk/sky130A/libs.tech/ngspice/r+c/res_typical__cap_typical.spice

.include /home/schippes/share/pdk/sky130A/libs.tech/ngspice/r+c/res_typical__cap_typical__lin.spice

* Special cells

.include /home/schippes/share/pdk/sky130A/libs.tech/ngspice/corners/tt/specialized_cells.spice

with this single change ngspice parsing goes from 30 seconds down to 6.5 seconds on my system. The real question is why ngspice (apparently) insists on parsing unused corners (@User also dealt with this issue a lot).

Thanks @User for jumping in and bringing the script to our attention. Thank you for the contribution, as well, these kinds of things are what take us forward on a daily basis - your work is much appreciated. I will take over the script and try it out in our environment - I am stuck on openlane now 🙂