ALF meeting, April 16, 2002, @ Monterey Design Systems, Sunnyvale CA Participants ============ Joe Daniels self Bhalchandra Thatte Sequence Alex Zamfirescu ASC Cho Moon Cadence Tak Young Monterey Wolfgang Roethig NEC Dennis Brophy Accellera Srinivas Maboosi Fujitsu Greg DuFour (phone) Mentor Kim Nguyen Tera Systems Kevin Grotjohn (phone) LSI Logic Plans, status and actions ========================= Next meeeting ------------- Conference calls, as need be in May and June. 15 July meeting @ Mentor in Portland needs to be rescheduled eventually, due to possible meeting conflict with JEITA. DAC publicity ------------- Overall DAC attendance possibly low. Therefore, no plan for BoF. Idea: create logo and CD, distribute CD @ Accellera breakfast, Put logo on poster of ALF-supporting EDA vendors IEEE status ----------- Initial ballot in May (next month!) too aggressive. August is more realistic. IEC --- IEEE std can become IEC std on fast track or by ballot, facilitated by Copyright agreement between IEEE and IEC. IEC is a mandate for industry certification especially in Europe. ALF will be presentedd to IEC next week in Austin, TX. Advice: Do IEEE and IEC ballot at the same time. Need to adjust ballot schedule for that. JEITA ----- JEITA = Japanese Electronic Industry Trade Association. Got invitation to participate in Accellera workshop in Tokyo, most likely in July. Preparing ALF tutorial for workshop. ALF publications ---------------- Showed part of DATE2002 paper and presentation. This is also available on ALF webpage. Another presentation will go online on VirtualEDACafe April 26. VirtualEDACafe is a virtual tradeshow with participation of Accellera and several EDA vendors. Informative discussion ---------------------- Explained similarities and differences between ALF, OLA, .lib, .lef Similarity with OLA: enables timing sign-off without SDF Difference to OLA: File format, not API Similarity with .lib, .lef, .tlf: file format, containing tables and equations Differences to .lib, .lef, .tlf: more comprehensive, vendor-neutral Also explained on the whiteboard: - extraction flow: related to ALF - OLA architecture: complementary to ALF - multi-voltage rail support: feature of ALF AI status --------- A.I. Design and test magazine - Alex Create one page, take picture from ALF presentation REOPENED A.I. Call Paul Menchini for ballot constituency - Wolfgang NEW A.I. Sign up country experts for ALF support in IEC meeting - Alex NEW A.I. Announce VirtualEDACafe tradeshow on ALF and OLA reflector - Wolfgang NEW A.I. round up IEEE members for participation in ballot (at least 3 per company) - all NEW Most old action items were resolved during the technical discussion. More new action items from technical discussion came up. A.I. Explain pitfalls, if INCLUDE statement is not properly used - Alex NEW A.I. schedule conf call for restrict_class - Wolfgang Desired participants: Kevin, Bhalachandra, Alex, Tim Ehrler Need to prepare a table comparing the feature sets of the different proposals NEW A.I. find character naming terminology used in unicode set or ISO??? - Alex NEW A.I. Propose a smart way to deal with smart quotes - Alex NEW A.I. send proposal for SI-units now for circulation to Paul - Wolfgang "SI" here stands for "Systeme international" (French), not signal integrity. NEW IEEE draft3 review ================== Chapter5 -------- Fig. 4: redundant lines between arithmetic model and arithmetic submodel Sec. 5.8: check whether the rule of declaration must precede reference is mentioned somewhere Book was not updated when compileing document (inconsistent pages and ToC) Chapter 6 --------- character terminology | = vertical bar ^ = "carat" or "hat" @ = ? # = "hash" or "pound" Required modifications within chapter 6: - Allow termination of in-line comment by end of file - Relational operator: use "less" instead of "lesser" - unary/binary is "arrity" - change "unsigned.unsigned" to "unsigned.[unsigned]" or "[unsigned].unsigned" - change "unsigned" to "unsigned_integer" Chapter 7 --------- Last paragraph in sec. 7.15 should be a comment, not normative text. Alex explained concerns about improper usage of INCLUDE statements. Group felt, that proper use is responsibility of user and environment, ALF standard should not attempt to prevent possible mistakes. Proposal for ALF header file ============================ Keywword with context-independent value set and default preferable to keyword with context-dependent value set and default. Therefore, change the following annotations: - change MEASUREMENT in "distance" context to DISTANCE_MEASUREMENT - change PURPOSE in "layer" context to LAYERTYPE - change PURPOSE in "array" context to ARRAYTYPE - introduce WIRETYPE in "wire" context instead of PURPOSE Work doc review =============== Acceptance of previously "closed" and "tentatively closed" items confirmed. Restrict_class (Chapter 12) --------------------------- Discussion postponed by request from Kevin. Non-ALF formats (Chapter 17) ---------------------------- Proposal: Do not overload INCLUDE statement, introduce ASSOCIATE statement. The semantics of the association shall be application-specific. ALF shall only provide a pointer for such applications supporting association. Example: CELL nand2x1 { ASSOCIATE "nand2x1.vhd" { FORMAT = vhdl; } ASSOCIATE "nand2x1.v" { FORMAT = verilog; } } Supplementary definition: KEYWORD FORMAT = single_value_annotation { VALUETYPE = identifier; CONTEXT = ASSOCIATE; VALUES { vhdl verilog c \c++ alf } DEFAULT = alf; } Note: more VALUES (e.g. OLA-related data) can be considered, if people ask for it. Region statement (Chapter 18) ----------------------------- Agreed to accept it, including BOOLEAN. Extended definition for Purpose annotation (Chapter 24) ------------------------------------------------------- superceded by proposal for ALF header file. Illegal statement (Chapter 25) ------------------------------ Accepted, pair it up with "nochange", similar as "setup" is paired with "hold" Control_polarity (Chapter 26) ----------------------------- Proposal contains 3 features: - change POLARITY to CONTROL_POLARITY, when used in conjunction w control signal, POLARITY is confined to single_value_annotation - eliminate composite signaltypes ***_control, ***_enable in cases where *** represents a mode appearing in the cONTROL_POLARITY statement - allow customized identifiers for modes Accepted after discussion. Eliminate driver cell and pin spec (Chapter 28) ----------------------------------------------- Explained the more powerful STRUCTURE statement on whiteboard. Accepted after discussion. Substitution of VIA reference (Chapter 29) ------------------------------------------ Substitute via reference by - via instantiation, if instance and/or geometric transformation required - annotation, if instance and/or geometric transformation not required Accepted after discussion. Submodels for routing direction (Chapter 30) -------------------------------------------- Issue: how to measure length for 45 degree routing. Proposal: median line Accepted after discussion. For all newly accepted items, a 2 week timeout period for objections is proposed. Acceptance is deemed firm, if no objection raised by May 1.