[comp.lsi] Circuits that break simulators wanted

hamm@cadillac.CAD.MCC.COM (Steve Hamm) (07/30/88)

A couple of us at MCC are working on a "different" circuit simulator.
Our problem is attempting to assure that "different" == "improved".
With our limited set of test cases, it is difficult to see whether we
are improving, say with respect to DC convergence, for a subclass of
circuits and/or fouling up another group of circuits.

So, could anyone with problem circuits (non-proprietary, please)
e-mail them to me?  I'd define problems as DC convergence, transient
convergence, timestep problems, or obvious wrong answers.  SPICE (or
common industrial variant) input style, using UCB standard models,
would be preferred.  Generally, the smaller the circuit the better,
although we could use a few large (1500+ transistor) examples as
well.  I cannot promise much in response other than thanks, but if a
circuit is not working due to some particularly trivial problem, I'd
be glad to mail a note back.

Assuming that I get any response, I'd be willing to share the test cases
I collect with anyone interested in either developing circuit simulation 
software or stress testing your current simulator.  Drop me a note.


-- 
Steven D. Hamm                | ARPA:  hamm@mcc.com 
MCC, VLSI CAD Program         | UUCP:  ...!cs.utexas.edu!milano!cadillac!hamm
3500 W. Balcones Center Dr.   | phone: (512) 343-0978
Austin, TX 78759