Liberty/ALF harmonization meeting, Dec 15, 2003 at Sequence, Santa Clara Participants ------------ Andrea Cosmin, Artisan Hemant Joshi, Artisan Jim McCanny, Cadence Matt Liberty, Cadence (phone) Joe Daniels, IEEE John Williams, Markanix Greg Hackney, Mentor (phone) Wolfgang Roethig, NEC Dave Allen, Sequence Kevin Kranen, Synopsys Maddu Karutanatne, V-cube Next meeting ------------ Mon, January 19, hosted by NEC, Santa Clara February meeting to be hosted by Artisan March meeting to be hosted by Cadence Review of DoD ------------- Ranking and discussion on "other items" in DoD, section 2.7. * Functionality - Kevin: We should focus on characterization item - Wolfgang: Important, as far as it is related to characterization - Matt: Sometimes timing needs to be infered from function e.g. flipflop, integrated-clock-gating-cell ... - Greg: should be covered and easy to translate Rank: 1.5 * DFT - Might be out of scope, since it is a very large topic. - We can bring it in later, if desired Rank: 3 * Types, attributes, basic stucture of library - Prerequisite for library description - Topics: Describing scope, unit, pintype ... Rank: 1 * Design rules (max cap ...) A lot of historical legacy for the semantics Could be covered as a timing subtopic Rank: 2 * Format extensibility - Harmonization of tool-specific library extensions - Allow vendors to represent proprietary information - Allow experimental exploration of new constructs - Define least common denominator Rank: 2, but some differences in opinion * Calculation methods Dave: Is the main scope translation or characterization? - Semantics of char. data crucial - Representation of data in table or polynom or other format is post-processing, not characterization - Requirement to have one characterization process, the results of which can be expressed consistently in liberty and ALF Rank: 1.5 Timing ------ Questions and issues brought up during review of the doc: Q(Maddu): Are there defaults for the timing-related attributes? A: Depends on the particular attribute. Q(Matt): How to specify internal pins in ALF? A: By declaring a pin with DIRECTION=none and VIEW=none. Q(Wolfgang): are the keywords intrinsic_* still relevant? A: probably not, Kevin to investigate. Did not find documentation for half_unate timing arcs in latest liberty doc, Kevin to investigate. Table 1.2 is incomplete, Wolfgang to add nochange_*_* items. Q(John): Does liberty support assertions, such as error messages? A: Liberty does not. ALF supports assertion in VIOLATION statement. Hemant: Semantics of slew_derate_from_library needs to be documented. Action item status ------------------ Write a proposal for ontology, suitable as a chapter in DoD Owner: Alex Zamfirescu Expected completion: TBD by Alex WIP Review the workdoc and give feedback, especially on ranking. Point out missing or misrepresented topics, if any Owner: all Expected completion time: per email before next meeting DONE - all "other" items are ranked now New action items ---------------- Write short proposal for generic libraries Owner: Alex Zamfirescu Expected completion: before next meeting OPEN Write short proposal for energy as independent characterization dimension Owner: John Williams Expected completion: before next meeting DONE - proposal posted, discussion defered to other meeting Make list of liberty items for which ALF representation is unknown Owner: Fereshteh Expected completion: before next meeting WIP - main areas: power, flow-specific representation Write short elaboration on requirements noise items (see DoD section 2.4) Owner: Matt Liberty Expected completion: before next meeting WIP - Cadence discussions ongoing Find out whether Sequence can give permission to use lib2alf as experimental testcase generator and distribute results to group Owner: Dave Allen Expected completion: before next meeting DONE - interested users should contact Dave Allen Write paragraph about rules for libray data usage model by application for possible inclusion in scope Owner: Alex Expected completion: before next meeting OPEN Write paragraph in DoD scope section saying what to do with necessary items for which liberty and/or ALF may not be ready yet. Owner: Wolfgang Expected completion: before next meeting DONE - included in DoD, end of section 2.1 Add following bullet items in DoD section 2.7 - Extensibility - Calculation methods Owner: Wolfgang Expected completion: before next meeting DONE - included in DoD, section 2.7 List interoperability with SDF as separate item in 2.2 (preliminary ranking = 1.5) Owner: Wolfgang Expected completion: before next meeting DONE - created section 2.2.2 Distribute new doc on timing Owner: Wolfgang Expected completion: as early as possible DONE - timing doc staged on web and reviewed in today's meeting Describe existence condition and value condition in ALF Owner: Wolfgang Expected completion: before next meeting DONE - see timing doc, section 1.3 Elaborate on SDF interoperability with liberty and ALF (e.g. mapping of SDF_COND etc to ALF) Owner: Wolfgang Expected completion: before next meeting DONE - see timing doc, section 2.0 Resolve possible incompatibility between timing type and actual timing data encountered in Artisan library Owner: Greg, Andrea Expected completion: before next meeting WIP - Arc exists for release of the preset and clear condition Find out whether semantics for net_capacitance, pin_capacitance etc. are documented in latest version of liberty user guide Owner: Greg Expected completion: before next meeting DONE - Semantics are currently undocumented Translate Greg's example with conditional timing arcs to ALF Use two methods: by hand and using lib2alf translator Owner: Wolfgang Expected completion: before next meeting DONE - examples staged on web. New AI ------ Investigate *_capacitance semantics Owner: Kevin Kranen Investigate relevance for intrinsic_* Owner: Kevin Kranen Give feedback on conditional timing doc and examples Owner: Greg Hackney Update DoD per today's discussion Owner: Wolfgang Update timing doc per today's discussion Owner: Wolfgang Prepare logistics for next meeting Owner: Wolfgang