Subject: Meeting Minutes for Dec 11, 2002
From: Amouroux, John (john_amouroux@mentorg.com)
Date: Tue Dec 10 2002 - 10:32:50 PST
SV-CC Meeting Minutes - Dec 11, 2002
General ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Rolling 12 Week Attendance List:
[xxxxx--------] Yatin Trivedi (ASIC Group)
[-xxx-x-------] Tarak Parikh (AT HDL)
[-xxxxx-xxxxxx] Francoise Martinole (Cadence)
[-xxxxx-x-xxxx] Stuart Swan (Cadence)
[--xxxxxxxx-xx] John Amouroux (Mentor)
[--x--xxx-----] Emerald Holzwarth (Mentor)
[-xxxxx-xxxxxx] John Stickley (Mentor)
[-------x-----] Duane Pryor (Mentor)
[--xxx-xxxxxxx] Doug Warmke (Mentor)
[-xxxxxxxxxx-x] Michael Rohleder (Motorola)
[xxxxx-xxxxxxx] Kevin Cameron (National Semi)
[-------------] Tayung Liu (Novas)
[x-xxxxxxxxxxx] Bassam Tabbara (Novas)
[-xxxxxx-xxxxx] Swapnajit Mittra (SGI)
[---xx--------] Darryl Parham (Sun)
[x------------] Simon Davidmann (Synopsys)
[x------------] Peter Flake (Synopsys)
[xxxxxxxxxxxxx] Joao Geada (Synopsys)
[xxxxxxxxxxxxx] Ghassan Khoory (Synopsys)
[x-xxxxxxxxxxx] Andrzej Litwiniuk (Synopsys)
[-xxxx-x------] Alain Reynaud (Tensilica)
[--xxx--x-----] Mike McNamara (Verisity)
[------x------] Kurt Takara (0-in)
[-------x-----] Dave Rich (?)
[-------xxxxxx] Joe Daniel (LRM Editor)
Last week's minutes approved
Swapnajit:
Got positive feedback from presentation last week to the general committee.
Kevin Cameron's Proposal :::::::::::::::::::::::::::::::::::::::::
Kevin: Main difference (from JohnS's proposal) is that John uses VPI for
access, while this one sets up and allocates everything on its own from data
obtained during elaboration, including context-specific C++ class calls,
even with multiple calls from within the same statement line. Context
contains instance-specific data.
Stuart: Is concerned because he feels that we're trying to model
instance-specific data on the C side, and is not sure if this is necessary
now.
JohnS: Then how does one call a context-specific function? This is what
we're trying to address here.
Joao: Does not want to invent new API's if VPI already has the capability.
Francoise: Technically there is no elaboration in Verilog, so how does
Kevin's proposal set things up? When are/how are the C names set up?
Kevin: C setup is done during the Verilog compile phase.
Swapnajit: We need to move onto the next agenda topic.
Kevin: We've just scratched the surface of this topic. But we need this to
be able to link C++. This proposal sovles this problem.
Joao: This solution makes the compiler become too aware of the C++ compiler.
Swapnajit: Please vote on this (1.1b) by Wednesday if you haven't already.
Please continue this discussion via email.
Andrzej's C Proposal ::::::::::::::::::::::::::::::::::::::::::
Swapnajit: Talked about delaying this second proposal to 3.2 if we cannot
finish this.
Francoise, Stuart & Others: We cannot delay this until 3.2 because our
contribution would be incomplete without it.
Andrzej: Prefers to keep natural layout that requires no overhead for
argument passing, including simple structures. Args will be passed by
reference, and vectors will be normalized.
JohnS: How about open arrays?
Andrzej: Parameters will be accessed via VPI.
Kevin: What if I want to pass by value?
Andrzej/JohnS: We already agreed against this. (Proposal 1, 10a & 10b)
Francoise: For example 1, can we directly modify o3?
JohnS/Kevin: The big quistion is whether or not to use VPI or some new
mechanism.
Andrzej: If you know the simulator's implementation, you can just use o3 as
a pointer. Otherwise you can use the indirect access functions.
Doug: Can we possibly define a standard layout for all the simulators for
4-state
Conclusion ::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Next meeting: Tomorrow, same starting time and same phone number, but we may
take two hours instead of one to try to finish up Andrzej's proposal.
Wednesday, Dec 11, 9AM-11AM PST
(877) 807-5706 (US toll free) or
(225) 383-8961 (International)
Passcode: 707131
This archive was generated by hypermail 2b28 : Tue Dec 10 2002 - 10:33:37 PST