Running two times ASSURA DRC hangs

[Nadine]

Hi all,


We have the problem that for some of us the second DRC run crashes
cadence.
I.e. it does no longer react to anything, not even redraw events. We
have to
"killall icfb.exe".
We're using (the unsupported) Fedora Core 3 on x86 and x86_64, Cadence
5.10.41
and Assura 3.1.4.
The pattern is that t's working just fine for about half of the people
here.
The other half usually gets the hang but not in 100% of the cases. It
seems
removing the last run's output before starting icfb improves things a
bit.

I've found that setting ASSURA_UI_NONMEMDATA to 1 might help according
to
http://groups.google.de/group/comp....61?q=assura+drc&rnum=7&hl=de#c0beda03871ad261
but it doesn't seem to, here.

Any other ideas?

Nadine

 

[Nadine]

Coincidence helped me here:
The problem is the flagUnconnectedVias option in the DRC rules for
UMC's 180nm
technology. Disabling this options fixes the crashes.

Nadine

 

[S. Badel]

thanks, i have had this problem too, but i lived with it.


stéphane

 

[Diva Physical Verification]

Coincidence helped me here:
The problem is the flagUnconnectedVias option in the DRC rules for
UMC's 180nm
technology. Disabling this options fixes the crashes.

Nadine

Is there any chance you could submit a SR for this problem? I have
forwarded your postings and they may be able to reproduce it with
something they have in-house, but a small testcase would help them be
sure they are fixing the problem you are seeing.

 

[Nadine]

Is there any chance you could submit a SR for this problem? I have
forwarded your postings and they may be able to reproduce it with
something they have in-house, but a small testcase would help them be
sure they are fixing the problem you are seeing.

I don't have much time at the moment because we need to get our chip
ready for next week's run. Just let me know by mid next week whether
you were able to reproduce this or not. I can prepare a small example
by then. But actually really every trivial cell triggers this here.
Just place an N_18_MM instance, no pins etc., and run DRC twice with
- flagOffGrid 0.01
- flagNon45
- flagAcuteAngle
- flagUnconnectedVias
and it hangs. The last of these options is the critical one. The others
are the typical set we commonly use. I guess it would also fail without
them.

Nadine

 

[Diva Physical Verification]

I don't have much time at the moment because we need to get our chip
ready for next week's run. Just let me know by mid next week whether
you were able to reproduce this or not. I can prepare a small example
by then. But actually really every trivial cell triggers this here.
Just place an N_18_MM instance, no pins etc., and run DRC twice with
- flagOffGrid 0.01
- flagNon45
- flagAcuteAngle
- flagUnconnectedVias
and it hangs. The last of these options is the critical one. The others
are the typical set we commonly use. I guess it would also fail without
them.

Nadine

I'll ask the Assura team if they can reproduce it, but you should assume
they can't and file the SR when you have the time. I'd try to do it
myself, but what I don't know about Assura is as great as what I DO know
about Diva. My being the Diva guy, and all that.