ALF DFT conference call, February 11, 11AM-12:30PM

Attendees:
Mike Andrews	Mentor Graphics
Greg Dufour	Mentor Graphics
Omar Kebichi	Mentor Graphics
Steve Pateras	Logicvision
Pierre Girouard	Logicvision
Tim Ayres	Synopsys
Wolfgang Roethig	NEC

Documents available for review:
* additions2alf2.0.february2000.preliminary.pdf (Wolfgang)
* alf4dft.05.pdf (Mike)
* summary of discussion items since last meeting (Greg)
* physical mapping examples (Logicvision)

Wolfgang's document had additions which reflected the state
of discussion since last meeting. The relevant chapters
for DFT were reviewed.

Chapter 1.5 Misc. amendments
----------------------------
Celltype is a functional classification, therefore PAD
will be taken out. PAD is a physical classification, using
the proposed keyword PLACEMENT_TYPE.
The proposal has been discussed in the physical workgroup
meeting, however, it is missing in the current spec.

A.I. put PLACEMENT_TYPE into spec. - Wolfgang

Celltype SPECIAL: maybe busholders should be in a different
category than fillcells and diodes. The functionality of busholders
is similar to like pull up, pull down and clamp cells, whereas
fillcells and diodes cannot be described using a FUNCTION statement.

A.I. DFT requirements for busholders etc. - Mike

Chapter 1.6 Items to be phased out
----------------------------------
SCAN - phased out

OFF_STATE - phased out

SCAN_USAGE - phased out

ENABLE_PIN - see OPERATION, need application note for substitution

A.I. Application note for ENABLE_PIN substitution - Wolfgang

POLARITY for output - phased out, need application note for substitution

A.I. Application note for output polarity substitution - Wolfgang

PRIMITIVE definition in FUNCTION - phased out

Chapter 2.5 SIGNALTYPE, OPERATION, SUPPLYTYPE
---------------------------------------------
A.I. refine the distinction between CONTROL, SELECT, ENABLE - Mike, Tim

composite signaltypes:
combination read | write with data | address is redundant

A.I. remove redundant combinations - Wolfgang

OPERATION triggered some discussion.
Wolfgang: The VECTOR in the context of OPERATION should contain the
sequence of input signals in conjunction with the expected output signals
to allow both test vector generation and verification.
Mike: The purpose of OPERATION should remain very clear. Still prefers
to have the read-off behavior defined seperately from OPERATION.
Tim: Predefined values for OPERATION (Mike's doc, proposed set comes from lpogicvision)
need more work.

A.I. More requirement analysis for OPERATION - all

Chapter 2.6 New usage models for CLASS
--------------------------------------
PURPOSE: Usage is for a VECTOR or a CLASS referenced in a VECTOR
to allow the application tool to identify the set of vectors relevant
for the tool. Need also PURPOSE=TEST.

PORTTYPE seems to be redundant, combination of SIGNAL_CLASS and SIGNALTYPE
allows to retrieve all necessary information for read port, writeport etc.

A.I. update DFT requirements, for example read-write-control using
     the SIGNAL_CLASS and OPERATION concept - Mike

General discussion
------------------
DFT tools need certain connectivity information. Do we need customized
attributes or properties for that? Wolfgang recommends to use the
CONNECT_CLASS and CONNECT_RULE construct instead of customized attributes
and properties, if possible.

A.I. CONNECT_RULE requirements for DFT - Pierre
A.I. application note for the above - Wolfgang

A.I. schedule next conf call - Wolfgang