Attendees:
Mark Hahn, Cadence (Chair)
Jin-sheng Shyr, Toshiba (Co-Chair)
Tom Dewey, Mentor Graphics
Steve Grouts, Sematech
Enrico Malavasi, Cadence
Dan Moritz, LSI Logic
Greg Schulte, Ambit
Jim Swift, IBM
New action items:
Who When What
---------- ------ --------
1. Jin, Jim 8/25 Discuss PVT-dependent constraints and relationship
to conceptual model
2. Steve 8/25 Update the parasitics boundary conditions
Open action items:
Who When What
---------- ------ --------
1. Greg 6/22 Write up an extended description of how tags work
-> 8/4
-> 8/18
-> 8/25
2. Jim 6/22 Write up an extended description of how operating
-> 7/21 conditions work
-> 8/4
-> 8/18
-> 8/25
3. Mark 8/18 Merge the taxonomy documents
-> 8/25
Completed action items:
Who When What
---------- ------ --------
1. Enrico 8/4 Write up a description of the "happiness"
-> 8/18 function for relating constraint priorities
2. Steve 8/4 Send email on die timing nomenclature
-> 8/18
Next Meeting:
The next meeting will be a face to face meeting on
Tuesday, 8/25/98, from 9-12 am PST.
Details:
1. Discuss taxonomy
We discussed specifying constraints in the HDL source
versus a separate file. Mark described the conclusions
from SC-WG, that generally specifying constraints in the
source file is undesirable. However, it can be convenient in
some cases where the design object to which the constraint
applies is unnamed, such as specifying the architecture for
the adder to be used for a particular "+" operator in the
source.
We talked about whether constraints might need to be a
function of environment parameters such as process, voltage,
and temperature. The idea here is that for a soft or firm
IP block, some constraints on the internal implementation
of the IP will generally depend on the design in which the
IP will be instantiated. As part of the conceptual model,
we need to define how an IP supplier describes the constraints
which are common to all implementations of the IP, as well
as the relationship between the environment and any environment-
dependent constraints for a particular implementation.
We also discussed a related topic, whether it was necessary or
appropriate to specify timing boundary conditions in the set
of constraints for an IP block. This is tied to the relationship
between constraints and models. One view of this is based on the
observation that a hard IP block will have a set of models, and
included in the models will be assertions which ensure that the
block will function properly in a particular environment. In an
expanded sense, the models for a soft or firm IP block could
include assertions which ensure that it is feasible to complete
the implementation of the block such that it will function properly
in a particular environment.
2. Presentation from Enrico on managing relative constraint priorities
Enrico presented an approach for allowing users to specify how
to make tradeoffs between multiple constraints. The group agreed
that this is an important capability, and that the happiness function
is an interesting way to support it. One potential issue with the
happiness function is that in the current form assumes there is a
single contiguous feasible region. If we want to support cases where
the solution space contains several disjoint feasible regions, an
extension to the happiness function would be required.
3. Agenda for face to face meeting
We talked about potential topics for the face to face meeting,
including
- review of timing boundary conditions taxonomy draft
- managing constraints when getting optimal performance from
tools requires tweaking the real constraints
- tag-based timing constraints
- operating conditions
- conceptual model
- schedule
Thanks,
Mark