( ESNUG 434 Item 10 ) ------------------------------------------- [11/18/04]
Subject: ( ESNUG 433 #8 ) PrimeTime-SI 2004.06 Bug Was An Odd Corner Case
> PrimeTime-SI's latest version 2004.06* seems to have a lot of stability
> issues. I'm having it crash on many normal situations like reporting
> timing from a restored session. This happens across multiple platforms
> for the same design that goes through without crashing on the earlier
> versions. I understand that 2004.06 is supposed to be better in terms
> of results (especially in terms of cross-talk delay computation) and bug
> fixes from the previous versions, but I wonder if anybody has encountered
> similar issues with this version.
>
> Is just me and my design or is it really PrimeTime-SI 2004.06?
>
> - Jay Pragasam
> Open-Silicon, Inc. Sunnyvale, CA
From: Jay Pragasam <jay.pragasam=user domain=open-silicon spot calm>
Hi John,
The crash that I had reported in the 2004.06* version of PTSI seems to
be real but it is confined to a spectrum of a specific design, run on
certain class of machines with certain kernels with a certain set of
libraries. After different combination of data exchanges with Synopsys,
I hear that they are able to get a similar crash though the stack trace
seems different. The issue is still under investigation as to what
exactly causes this crash. It is a little strange that none of the other
designs here crash with the 2004.06* versions, whereas this one design
crashes on all the architectures. Since this seems to be a cornercase,
we are moving forward using 2004.06-SP1-1 in different projects with
caution but are expecting Synopsys to root cause this issue before the
case is closed.
- Jay Pragasam
Open-Silicon Inc. Sunnyvale, CA
---- ---- ---- ---- ---- ---- ----
From: Uyen Tran <uyen.tran=user domain=synopsys spot calm>
Hi John,
PrimeTime SI 2004.06 is a stable release and is being used successfully by
many of our customers. This problem seems to be isolated to OpenSilicon's
design and usage of the restore_session feature, which was introduced in
2003.12. We have not seen this type of problem from other customers.
However, every time there is a fatal in PrimeTime, Synopsys takes this very
seriously. We have been working with OpenSilicon to reproduce the problem,
but it has taken time due to lack of access to third party libraries.
We have recently received the complete testcase and are working to identify
the root cause of the fatal. We are confident that we will be able address
this issue shortly.
- Uyen Tran
Synopsys, Inc. Mountain View, CA
|
