Subject: Draft Nov. 5, 2002 meeting minutes
From: Amouroux, John (john_amouroux@mentorg.com)
Date: Tue Nov 05 2002 - 12:47:07 PST
Meeting minutes for the SV-CC Committee
November 5, 2002, 9:00-10:00am PST
Attendees:
[-xxxxxxx--] Yatin Trivedi (ASIC Group, Ex-Chair)
[----xxx-x-] Tarak Parikh (@HDL)
[-xx-xxxxx-] Francoise Martinole (Cadence)
[xx--xxxxx-] Stuart Swan (Cadence)
[-----xxxxx] John Amouroux (Mentor)
[-----x--xx] Emerald Holzwarth (Mentor)
[----xxxxx-] John Stickley (Mentor)
[-----xxx-x] Doug Warmke (Mentor)
[-xx-xxxxxx] Michael Rohleder (Motorola)
[xx-xxxxx-x] Kevin Cameron (National Semi)
[----------] Tayung Liu (Novas)
[xxxx-xxxxx] Bassam Tabbara (Novas)
[----xxxxxx] Swapnajit Mittra (SGI, Chair)
[-x----xx--] Darryl Parham (Sun)
[xx-x------] Simon Davidmann (Synopsys)
[x-xx------] Peter Flake (Synopsys)
[xxxxxxxxxx] Joao Geada (Synopsys)
[xx-xxxxxxx] Ghassan Khoory (Synopsys, Co-Chair)
[xxxx-xxxxx] Andrzej Litwiniuk (Synopsys)
[--x-xxxx-x] Alain Reynaud (Tensilica)
[-----xxx--] Mike McNamara (Verisity)
[---------x] Kurt Takara (0-In)
1. Swapnajit's general issues
Yatin has stepped down from the committee chair position, and Swapnajit is
assuming this responsibility.
Swapnajit forsees that with the many issues facing us we will need to
determine a prioritization scheme. He will be proposing something in a
future meeting, along with some rules to determine the lifetime of issues.
Swapnajit will not be present at the Nov. 7th face-to-face meeting, but he
is working with Bassam (who will chair this meeting) and all of us to get
the agenda and some basic slides together.
Swapnajit is still working with the folks that manage the eda.org server to
try to get the issue pages moved over there to eliminate the browser
pop-ups.
2. Roll call
The "rolling" attendance list at the top of the meeting notes the attendees.
3. Approval of minutes
Last week's minutes were approved.
4. Discussion of ISSUE 1.4
ISSUE 1.4, titled "No clear relationship to other APIs"
In this issue, Michael has pointed out that if Direct C code calls other API
functions (notably PLI functions), then the simulator would have to ensure
that objects like internally-cached data and the like would be flushed to
provide correct functionality to the model. But in doing so, the simulation
may slow down by either having to perform such steps or by limiting
optimization steps.
Andrzej proposed that maybe only access should be granted to those objects
on the formal parameter list, and Kevin noted that with C++ we have the
"const" paradigm. Kevin also pointed out that SystemVerilog already has
facilities for such things.
Michael and Doug both clarified this issue to be that we simply need to
decide if a C routine can call PLI-like routines or not. And if so, how do
we also allow for optimizations?
Joao noted that simple PLI functions like io_printf will be very useful for
C routines to call, and others noted that allowing the PLI calls will makes
things much easier for the users. Joao proposed adorning the safe (no PLI
calls) vs. unsafe (makes PLI calls) functions with something like "extern
"D", for dirty/unsafe functions. He noted that we could make this adornment
flexible much like the attribute list in the "C" fopen syntax. Michael like
this idea, but just wanted a simple syntax.
Andrzej proposed a different method, where users would call a special
routine to tell the simulator that they will now be getting dirty, but after
some discussion the group consensus was that a simple "extern D" type of
syntax was preferered.
ACTION: Joao, Michael, and Doug to discuss this syntax off-line.
5. Discussion of ISSUE 1.5
ISSUE 1.5, titled "Should use a common and unique prefix for all functions"
All quickly agreed that a standard here was good, and required. Swapnajit
suggested that we vote on this at Thursday's meeting, so we will need some
pre-meeting proposals to be floated around beforehand so that we can all
make informed votes.
ACTION: All/anyone propose a prefix scheme. At the meeting the prefixes
"sv_" and "vc_" were mentioned.
6. Discussion of agenda for the Nov. 7th meeting.
Some ideas were:
C calls Verilog
Verilog calls C
DirectC-style vs. System C-style syntax
Cadence's C++ proposal
Joao and John S's HDL to C communication ideas
Andrzej's proposal
Andrzej's [to be filed] ISSUE about C calling Verilog system tasks
ACTION: All please send your items to Swapnajit so he can finalize the
agenda.
The meeting will be held at:
Synopsys
700 East Middlefield Road
Building B, Diamond Conference Room
Mountain View, CA
9am - 5pm PST
[Teleconference numbers will be provided later.]
7. Meeting adjourned.
[These minutes were taken by John Amouroux. Please send any correction
requests to mailto:johnny@model.com.]
This archive was generated by hypermail 2b28 : Tue Nov 05 2002 - 12:47:36 PST