Is there an easy way to save off the results of a ...
# xycer
Robert Rogers
10/22/2021, 6:08 PMIs there an easy way to save off the results of a netlist parse/problem setup prior to doing a transient sim?
e
Eric Keiter
10/25/2021, 6:32 PMYou can do “Xyce -syntax netlist” and Xyce will just do a syntax check and then exit without simulating.
Eric Keiter
10/25/2021, 6:32 PMYou can also do “Xyce -norun netlist” and Xyce will go a little farther, checking syntax and topology before exiting.
Eric Keiter
10/25/2021, 6:33 PMThere is also “Xyce -count netlist” which will stop earlier than -syntax or -norun. It just does a count of devices and then exits, very early.
r
Robert Rogers
10/25/2021, 6:44 PMI was more interested in saving off whatever the representation of the system is at the point just prior to initiating the numerical analysis.
e
Eric Keiter
10/25/2021, 6:45 PMAh, I see. At the moment, no, there isn’t a way to do that. We’re in the process of writing a new parser that should have that capability, but it will be a while before that is ready.
Eric Keiter
10/25/2021, 6:46 PMThere is some parallelism in the parser, so you can reduce the parse time (somewhat) by running in parallel. BTW
r
Robert Rogers
10/25/2021, 6:48 PMUnderstood! Right now I'm trouble shooting some weirdness with my openmpi install, I'm observing the solver randomly hanging during the parsing step for reasons we don't understand (when ran in parallel) I'll be sure to submit a bug report if we can isolate the issue
e
Eric Keiter
10/25/2021, 6:49 PMOK, thanks. Yes, if that is happening, we’d want to see the bug report when you can isolate it.
r
Robert Rogers
10/25/2021, 6:50 PMI ran the "is your mpi install broken?" check you guys included with the source. Nothing doing there
Robert Rogers
10/25/2021, 6:51 PMSo I've had to go down into the weeds.
e
Eric Keiter
10/25/2021, 6:51 PMIs this a netlist that parses fine in serial?
r
Robert Rogers
10/25/2021, 6:51 PMYep! And inconsistently when ran in parallel
e
Eric Keiter
10/25/2021, 6:52 PMHmmm. Is it a netlist you can share? (I understand wanting to create a small test case that reproduces the problem)
r
Robert Rogers
10/25/2021, 6:53 PMUnfortunately no. Part of the process I'm going through now includes generating a sharable netlist that reproduces the problem.
e
Eric Keiter
10/25/2021, 6:54 PMAh. I understand. Thanks!
r
Robert Rogers
10/25/2021, 6:55 PMIt might not be possible. The netlist exhibiting the behavior involves TSMC's 12LP PDK, so if it's due to something proprietary in that we'll be out of luck
e
Eric Keiter
10/25/2021, 6:55 PMAh, OK.
r
Robert Rogers
10/25/2021, 6:56 PMIf I can manage it with sky130, I'll be sure to share.
e
Eric Keiter
10/25/2021, 6:56 PMgreat
2 Views