ALF meeting April 20, 2000 at Philips

Attendees:

Jay Abraham		SiliconMetrics
Shir-Shen Chang		Synopsys
Mike Andrews		Mentor Graphics
Kent Moffat		Mentor Graphics
Dan Moritz		LSI Logic
John Peters		Philips
Joe Daniels		Intrinsix
Wolfgang Roethig	NEC

ALF 2.0 release schedule
------------------------

Plan to have a first round of official spec. review in May.
We signed up the following reviewers so far.

Jay Abraham (SiliconMetrics) -> delegate to someone
Shir-Shen Chang	(Synopsys) -> delegate to Tim Ayers
Mike Andrews (Mentor Graphics)
Dan Moritz (LSI Logic) -> delegate to someone
John Peters & Tim Ehrler (Philips)
Paul Zukowski (IBM) for physical ALF only

Wolfgang will call for more reviewers by email.

Spec. review
------------

- CELLTYPE "SPECIAL"
Mike: need busholders and other definitions for special cells.
Proposal: introduce new ATTRIBUTE for cell with celltype SPECIAL
BUSHOLDER | FILLCELL | CLAMP | <electical_component>
<electrical_component> = DIODE | CAPACITOR | RESISTOR | INDUCTOR }
CLAMP is a cell with pin(s) tied to a fixed logic value.
Value and strenght can be found in the BEHAVIOR description.

- CELLTYPE "BLOCK"
keep CELLTYPE BLOCK as well as PLACEMENT_TYPE BLOCK,
since both aspects are well-defined.

- CELLTYPE "CORE"
complex element which can be represented in various ways,
e.g. netlist, partial netlist, black-box, grey-box ...
can contain PRIMITIVE or behavioral code etc.
Introduce attributes PLL, DSP, CPU for cores

- SCAN_USAGE to be phased out
o.k. is confirmed by Mike.

Explanation why no default for EDGE_NUMBER in PERIOD:
Sequence of events is periodic, no matter from which edge
the period is measured (maybe a picture will help).

- mode-specific POLARITY o.k.

- SUPPLYTYPE: change BIAS into REFERENCE

Shir-Shen: Need apps note how to model Flipflop with "load",
using ENABLE and ACTION. Maybe we should introduce LOAD_ENABLE
and LOAD_CONTROL.

- BITMAP
Feedback from Mentor: BITMAP may be lengthy for large memories.
Put BANK back in spec. and include an example.
Also present alternative proposal:
Description of BEHAVIOR engine capable of translating the
test algorithm into valid test signals. This proposal is
more general than BITMAP, but maybe an overkill.

A.I. update spec. according to notes above - Wolfgang