SV-EC Meeting Minutes 5 August 2002 (--a-) Adam Krolnik (LSI Logic) (---a) Alec Stanculescu (Fintronic) (a-aa) Cliff Cummings (Sunburst) (-aaa) Dave Kelf (Co-Design) (aaaa) David Smith (Synopsys) (-aaa) Dennis Brophy (ModelTech) (---a) Francoise Martinolle (Cadence) (aa-a) Heath Chambers (HMC) (aaaa) Karen Pieper (Synopsys) (aaa-) Kevin Cameron (National) (---a) Kurt Takara (0-in) (--a-) Michael McNamara (Verisity) (aaaa) Mehdi Mohtashemi (Synopsys) (-aaa) Neil Korpusik (Sun) (aaa-) Paul Graham (Cadence) (aapa) Peter Flake (Co-Design) (a---) Roy Armoni (Intel) (aapa) Simon Davidmann (Co-Design) (-aaa) Stefen Boyd (Boyd Technology, Inc.) (aa--) Steven Sharp (Cadence) (-aaa) Tom Fitzpatrick (Co-Design) a => Attended p => Attended by proxy - => Missed Action Items: 1. Kevin to add more detail to proposal for "Dynamic Process Control" (left over from previous mtg) 2. Mehdi to verify that error produced for 2d array of enum. 3. All proposals for top 6 items must be in by 15 September. 4. CC to develop FSM slides. 5. CC and PF to discuss FSM enhancements and implied register types. 6. DS to prepare meeting notice and agenda for 4 September full-day meeting in Mountain View at Synopsys building. 7. DS to review with VG when ownership of donated material transfers to Accellera. 8. Fitz to verify that enums auto-cast to integers when passed to function. Similar to example on p. 1-14 of Vera doc. 9. MM to find out the maximum bit width of a bit-vector used to index an assoc array. Tom Fitzpatrick is taking the minutes (edited by David Smith for distribution). Discussion: 1. Review of minutes DS: Asked for approval of minutes from last meeting NK: p. 1-10 enum doesn't specify if enum values start at 0 or 1. Please ammend minutes to reflect this AR for Synopsys to clarify. TF: Move to accept minutes as modified. AS: Second Passed unanimously 2. Review Action Items Kevin not available to discuss process management enhancement. Mehdi distributed word document with clarifications on his ARs from last meeting. Discussed later in meeting. Tom Fitzpatrick (later in the meeting) commented on his AR that SV is able to auto-cast enums to ints when passed as 2d arrays (page 1-14 of donation). 3. Prioritization of open 12 items. DS claimed to have distributed the summary of the prioritization votes received thus far. (It was available by the end of the discussion on the reflector). No one had received it thus far. DS would like to choose a cutoff point on the prioritized list for which we can get proposals in to be 9/15/02. Alex will make proposal for external modules. Top 6 priorities: alias process control data channels FSM extern module inferred reg type 4. Discussion on top 6 items. DK asked whether data channels is a new proposal or if it is a discussion on how interfaces can meet the data channel requirements. DS said that if interfaces can handle the requirements, then a new proposal is not needed. CC questions on proposals. First is the alias. SB: this began with a discussion of "jumpered ports" from ANSI-style port lists CC: Alias construct would allow name mapping to assist in .* port list. Is this what is meant by the alias proposal? AS: The alias is not simply part of the .* usage, but has other uses for. TF: Two aspects of alias. One is to give different names to the same object AS: Generalization of a name to be equivalent to a slice of some other item FM: VHDL has an alias which allows aliasing to any subelement of a static element - anything with a declared identifier. DS: this is discussed in section 4.3.3 of the VHDL LRM. TF: The other aspect of alias is a name-mapping algorithm for .* CC: Will start putting together a proposal. FM: Are we proposing aliases to hierarchical references? SD: Can already be done with 'define or via direct assignment. Allowing aliases to heirarchical references will cause problems if/when the hierarchy changes. This is part of the value of interfaces. AS: Hier. refs should be recommended not to be used for this, but since some people like them, we should not disallow them. They are not safe. SD: Assignments in $root can be referenced from anyplace in the design. CC: Will not include hierarchical refs in the original proposal. Next was a discussion on FSMs. CC: Most interested in different enum types in FSMs. Question about hierarchical/concurrent FSMs. Cliff still waiting for someone to champion the hier/conc FSMs. DS: Donation from Co-Design included hierarchical FSMs, but not multi-clock FSMs, which are still waiting for a champion. CC: What if we add enum_onehot type? Would make it easier to change between "regular" FSM and one-hot FSM encoding instead of having to change the code. This proposal would be an extension of the original donation, but separate from the hier/conc FSM discussion. Action: CC to put slides together. DS: The original donation of FSMs fell down the list below the cutoff point. Last discussion was on implied register types. Action: CC and PF to discuss FSM enhancements and implied register types 5. Discussion on Synopsys support for donation review. DS: Sent out email that Synopsys is willing to perform additional analysis of the donation (7/31/02, "Agenda for Monday's Meeting" email) MM: Synopsys wants to be more proactive in helping get through the donation. Requests a forum to discuss the testbench donation in more detail. Would like to do this in the next two weeks, having the next meeting face-to-face hosted at Synopsys. DK: This would be useful, but we need more notice. DS: Proposed the week of 8/26-8/30 SD: Please distribute slides ahead of time so that folks can attend via phone. FM: Please describe the scope of the meeting MM: Going through the entire scope of the donated document. Addressing inconsistencies and answering questions. General discussion about whether it could be done via 3-way teleconference between California, Boston, and UK. Other options are to have it after Labor day. Shooting for 9/4. CC: Face-to-face meeting might facilitate additional donations DS: VG has deadline of mid-September for non-competing donations to be received. So far, no one has announced an impending donation. Action: DS to review with VG when ownership of donated material transfers to Accellera. SD: Will we be voting on the donated Vera document, or on potential "children" of this document? DS/CC: The face-to-face meeting is a chance for Synopsys to rework and clarify the donated document to facilitate the acceptance. 6. Review Synopsys response to questions (Mehdi's action items from previous meeting). Reviewing MM's email (word doc SV_EC_tb_questios.July22.doc) attempting to answer questions that came up from 7/22 meeting. Action: Fitz to verify that enums auto-cast to integers when passed to function. Similar to example on p. 1-14 of Vera doc. Discussion: How does the task of analyzing the donation fit vs. the other 12 items on the priority list? The donation analysis is a separate task. The committee will consider the donation in parallel (half-time) with considering the issues list. After the donation is reviewed and (if) accepted, each piece of the donation will be assigned to the appropriate issue on the issue list or scheduled as appropriate. SD: Can you use the contents of a structure as the index to an assoc array? MM: in VeraLite, assoc arrays are restricted to allow only integers or strings as the index. MM: Cannot use string to index an associative array. Action: MM to find out the maximum bit width of a bit-vector used to index an assoc array. 7. Next Meeting: 8/19, 11am-1pm PDT Dial-in info: The Participant code you will need to enter is: 516134 The phone numbers are: Toll Free Dial In Number: (877) 233-7845 International Access/Caller Paid Dial In Number: (505) 766-5458 8. Meeting adjourned.