SV-EC Meeting Minutes 28 June 2002 The meeting was jointly run between SV-BC and SV-EC. We covered the SV-EC agenda first. Attendees consisted of: (a) Cliff Cummings (Sunburst) * (a) Heath Chambers (HMC) * (a) Karen Pieper (Synopsys) * (a) Kevin Cameron (National) * (a) Mehdi Mohtashemi (Synopsys) * (a) Paul Graham (Cadence) * (a) Peter Flake (Co-Design) * (a) Rog Armoni (Intel) * (a) Simon Davidmann (Co-Design) * (a) Steven Sharp (Cadence) * * indicates elligible to vote on consensus issues The remainder of these minutes are for the SV-EC only. Action Items: 1. Send a new copy to the reflector and describe the subscription process. 2. Review/Approve these minutes at the next meeting. 3. Vote on the mission statement in section 2 of the attached Operating Guidelines at the next meeting. 4. Vote on the process defined in sections 4 A and B of the attached Operating Guidelines at the next meeting. 5. Each member review the items in section 4A and volunteer to generate proposals as appropriate. 6. Next meeting will start the review of the first two chapters of the donation. Descussion: 1. The first order of business was to review and vote on the committee voting policy. Cliff Cummings made the suggestion to include support for 75% overall attendance in addition to the last 3 out of 4 rule. Based on the changes and wording in section 4C of the attached Operating Guidelines a vote was taken and passed. Cliff will document the vote in the SV-BC minutes. 2. The next order of business was to announce that Stephen Boyd has accepted the co-chair position. 3. The email reflector (with instructions for subscribing) was discussed (with incorrect instructions from the chair). The correct instructions have been added to Section 1 in the "Operating Guidelines" attached. The next order of business was to review the Committee mission. Simon Davidmann raised the issue question of whether the mission was to create one language or two languages. He made the observation that Vera was a consistent language that worked well but it had syntax and semantics that are not in Verilog. David Smith observed that this was also true about the ESS donation based on SuperLog and that the required semantics and syntax was added to the language. Simon responded that he believes this to be technically wrong. The facts are: SUPERLOG was designed as an extension to Verilog syntax and semantics and is completely consistent with it. When ESS and DAS were donated to Accelera they both were already completely compatible with Verilog 95, 2001 syntax. A fundamental strategy of the language design was to extend Verilog, not be an alternative to it. There were no modifications to ESS from donation to standardization as System Verilog - just deferral of some items, and the addition of one. No changes to syntax or semantics. Vera was not designed as an extension to Verilog, but as an extra language, an HVL - that has constructs to connect one language to another. I believe we should be looking to focus on Vera as an HVL - in the same way that Verilog95 was standardized it was not modified by OVI to break all the existing Verilog code that people had working with it. We don't want to start modifying Vera - it already works - we need to keep it consistent with the legacy code that is out there - otherwise we would be doing a disservice to its users and Accellera members. A further observation by Simon was with respect to the "cycle based" nature of the testbench and the lack of cycle based semantics within Verilog. Kevin Cameron observed that cycle based semantics are a sub-set of the Verilog specification. The result of the discussion was the addition of an additonal statement in the Mission Statement that we will use a single syntax. There was also an extended discussion on whether we are generating one LRM or two and whether this means one language or two. David Smith suggested that we delay the decision on the LRM until after we review the Testbench donation. The suggested change to the mission is added in 2B of the attached Operating Guidelines. 4. The process to be used by the company was discussed. The meeting schedule was reviewed and Karen Pieper observed that 2 September is a holiday. We adjusted the schedule by skipping this day. The modification is in section 4 of the attached Operating Guidelines. The process to be used for reviewing the outstanding list of extension items was discussed. Simon Davidmann observed that many of the items could be research projects in their own right. He also suggested that we should decide not to handle some of them. David Smith indicated that the suggested process would provide a first level of pruning since if there was no one interested in spending effort then we do not have to address the extension. In addition, once the proposal is made then we can spend the time to decide if we want to take it through to completion based on the proposal. Based on the suggested process Kevin Cameron and Cliff Cummings volunteered to generate proposals. This is documented in section 4A of the attached Operating Guidelines. The discussion of the donations started with the suggestion by Simon that we should generate a set of requirements for the testbench extension and then solicite donations from companies. David Smith indicated that this is not the process for donations that has been established by the board and TCC chair. A discussion proceeded where Simon provided information that the board is currently reviewing the donation process and may make some changes. David observed that until the TCC chair or the board communicates procedural changes we need to follow the currently established operating policies. The discussion then moved on to a review of the policy to be used for processing the Testbench donation. Simon suggested that we should vote at the next meeting because, if we expect other donations, we will not have enough time to analyze and evaluate the donations in the required timeframe. David Smith indicated that until there are more donations it would be beneficial to take the same care with the Testbench donation that we took with the ESS donation before voting on it. 5. The working milestones were discussed with the only change being the delay of the 2 Sept meeting until 16 Sept. 6. Final discussion was over the action items and votes planned for the next meeting. See the notes above. ============================================================================== Operating Guidelines: 1. Communication The email reflector is: sv-ec@eda.org To subscribe please add the following line to the body of a text (non- html or rtf) email message: subscribe sv-ec and send to: majordomo@eda.org The website for the committe is: website: http://www.eda.org/vlog-pp/sv-ec 2. Committee Mission A. Review SV-Enhacenment Committee formation definition Focus on the enhancement and focused area of donations. The only new donation we have is Synopsys. In addition to the committee list we have the TestBench from Synopsys. B. Establish Mission (based on the formation definition) Development System Verilog 3.1 LRM based on clean-up and extensions. Process any donations provided to the committee by the TCC chair. Extend the Verilog language to system design and for testbench using a single syntax. May be modified Based on discussion during call. 3. Goals/Objectives A. Resolve the open list of System Verilog 3.1 extensions. B. Process the Synopsys VeraLite TestBench donation C. Rationalize the the TestBench extensions with System Verilog as appropriate 4. Process Meeting schedule: Every 2 weeks starting 8 July 2002 from 11:00am (PDT) to 1:00pm. 8, 22 July 5, 19 August 16, 30 September 14, 28 October 11, 25 November 9, 16 December 6, 20 January 3, 17 February The basic process is to review, evaluate, vote, and then LRM development for all activities. A. Item List For each of the listed extensions: i. Find advocate (today!) and assign milestone for completion ii. Advocate develops proposal that includes: Problem statement - requirements to be met Analysis - Describe issues raised by the requirement and interactions with the 3.0 LRM Solution statement - Describe syntax, semantics, and examples for the proposed solution iii. SV-EC reviews proposal and votes to proceed or table iv. If vote is to proceed then develop LRM modifications, including BNF. v. Committee votes on LRM modifications Consolidate all SV-BC and SV_EC LRM modifications into single document Committe votes on overal LRM to submit to parent committee/board Items in extension list are: i. Data channels ii. Pointers (Kevin C.) iii. Force/release with strengths iv. FSM (original ESS donation) (Cliff C.) v. Extern modules vi. OO (Kevin C.) vii. Data path (if donated by Cadence) viii. Alias (Cliff C., Kevin C.) ix. Inherited declarations x. Multi-clock FSM xi. Inferred reg types (Cliff C., Kevin C.) B. Donation Must analyze, evaluate, and accept or reject the donation through a formal vote. Draft LRM after donation accpetance by technical committee. i. Review all 8 chapters of donation. Purpose is to analyze and evaluate donation. During review develop list of issues organized by: a. missing information b. inconsistent information within donation c. conflicts with Verilog/System Verilog (both syntax and semantics) d. items to be migrated to System Verilog ii. Resolve issues in categories a and b. iii. Vote on acceptance or rejection of the donation iv. Resolve issues in category c and migrate issues in category d to extension list (as appropriate) v. Create LRM based on initial donation that follows the IEEE Verilog style including BNF vi. Committe votes on overal LRM to submit to parent committee/board C. Voting structure and rules 3 out of the last 4 meetings must be attended or 75% overall attendance (this is to handle the first three meetings) to be eligible to vote. Vote on technical issues will be a simple majority of all attendees (not limited to one per company) All procedural and final (accept donation, LRM) approvals (other items at the chair's discretion) will follow the one vote peru company rule. Each company has a designate (with proxy support). IEEE members, who are not an Accellera member or commercial affiliation will have an indivdual vote as well. Subject to TCC chair change/approval 5. Milestones A. Review external milestones September 16 2002 (4 meetings from now): Each issue item should have an assigned owner. The owner must develop changes/proposed enhancement and obtain sub-committee approval. Donations should have been analyzed, evaluated, and voting complete. December 2002 (7 meetings after first milestone) Draft of System Verilog 3.1 LRM (with SV-BC collaboration) Draft of TestBench LRM (may be part of 3.1 LRM) February 15 2003 (5 meetings after first draft) Final draft version to the Accellera board B. Establish committee milestones 8 July Chapters 1 and 2 of donation Lexical, types, operators, expressions Subroutines (any extension proposals scheduled) 22 July Chapters 3 and 4 of donation Sequential control Concurrency (any extension proposals scheduled) 5 August Chapters 5 and 6 of donation Interfaces to DUT Signal Operations (any extension proposals scheduled) 19 August Chapters 7 and 8 of donation Class Linked Lists (any extension proposals scheduled) Vote on donation and all completed proposals 16 September - December Milestones TBD for LRM creation 6. Deliverables System Verilog 3.1 LRM TestBench LRM