SV-EC Committee Meeting. Monday March 5 2007 11:00am - 1:00pm PST [Minutes distributed for review, to be approved at next meeting] (121202020102020) Day (481593604882505) (000011111100000) Month (889900112211223) (000000000000000) Year (666666666677777) --------- Attendees ---------- (-AAAAAAAAAAAAAA) Arturo Salz (--AAA-AAAAAAA-A) Cliff Cummings (AAAAAAA-AAAAAAA) Dave Rich (AA-A-AAA-AAAAAA) Francoise Martinolle (-AAAAAAAAAAAAAA) Mehdi Mohtashemi (AAAAAAAAAAAAAAA) Neil Korpusik (AAAAAAAAAA-AAAA) Ray Ryan (AAAAAAAAAAAA-AA) Gordon Vreugdenhil (AAAAAA--AAAAA-A) Steven Sharp (--AAAA-A-------) Phil Moorby (---AA-AAA-AAAA-) Doug Warmke (AAAAAAA---AA-A-) Stu Sutherland (-AAAA--AAAA-A-A) Heath Chambers (-AAAAAA-A----AA) Don Mills (--AA--A---A-AAA) Jonathan Bromley (--A------------) Logie Ramachandran (----AAAA---A---) Melvin Cardoza (-----A-AAAAAA-A) Mark Hartoog (-------A-------) Satia (from Intel) (--------AAA----) Rob Slater (Freescale) (-------------A-) Alex Gram (Mentor) ^ |------- non-voting meeting ** Minutes taken by Neil Korpusik and Mehdi Mohtashemi ////////////////// March 5, 2007 ///////////////////////// Agenda: 1. IEEE patent policy ref: http://standards.ieee.org/board/pat/pat-slideset.ppt Move: Dave - Assume that the patent policy was read Second: Neil Abstain: none Opposed: none passed 2. Review meeting minutes/Notes: --------------------------------------------------- http://www.eda.org/sv-ec/Minutes/SV-EC_Meeting_February_20_2007_Minutes.txt 3. P1800 meeting: --------------------------------- - Draft 1 : 150 mantis items 11 were flagged as having issues with the edits (3 from svec) These will be fixed in the merged LRM since it will be out soon. - Merged LRMs ECD 3/15 (Stu might be late) - he will send something on 3/15 Will contain line numbers down the side. BNF snippets - Stu will check if the IEEE will allow to be informative - SVEC - Mehdi gave a brief summary of the proposal for mantis 890 - SVAC - Dmitry's list of proposals AI:Neil - send email on 890 to ac,bc,cc (post Cliff's portion on mantis) Gord - wants initial review of the svac proposals (in svbc) AI:Neil - forward Dmitry's slides to the active set of people on svec. (get them from Dennis) 4. Continue review and discussion on Mantis 890 and related mantis items ------------------------------------------------------------------------- a) Rob's use model - not discussed. b) 890 08-31-06 clarifications in program and clocking blocks (Doug) Cliff sent out an update. Cliff - made the changes that Neil sent to Cliff (from f2f). - the Pre-Observed region was the biggest change. Gord - should say "read-only". Use similar wording to 9.3.2.9 (Postponed). Neil: part1 comments. section 15-10, when use as word as should be removed Francoise - Stu asked for wording clarifications from svcc (no response yet) Gord - we must add more wording since there is no feedback to Active - concerned about the simulators obligation in Pre-Observed Cliff - "Within this region it is illegal to write values to any net or var or to schedule any events in the current time slot." Arturo - there is an even newer version. version 9. He sent the latest version to Neil (to see if there is overlap with the issues that Neil has noticed). Dave - there are still parenthetical comments (that are really normative text) - text left in parens. Gord - at least one place where a parenthetical comment is used (i.e. in the newest version). Cliff - continuous assignments are ok in program blocking assignments and non-blocking assignments in program ok process events from a region -- cliff is happy with the way things are turning out. AI:Neil - send feedback to Doug on any remaining outstanding issues with 890. - ask for feedback by Wed of next week from other committees. - add Cliff's writeup to mantis 890 (part 2) - have other committees review what is already on mantis from Doug. 236 - should be resolved... ##0 Move: Jonathan : close 236 (covered by 890) Second: Neil Abstain: none Opposed: none passed closed covered by 890 608 - being clarified. Move: Jonathan : close 608 (covered by 890) Second: Steven Abstain: none Opposed: none passed closed covered by 890 609 - association of clocking block, need a syntactic form. Jonathan: it may not be completely covered in the 890. the mantis presents a point of view - ## on LHS is now more rigorously defined ## on RHS is still a different usage. Arturo - referred to ## as a statement. linked to 890 (decided to keep the two forms of ##) Move: Jonathan - move to close 609 ("hat trick") Second: Gord Abstain: none Opposed: none passed closed linked to 890 564 - Cross-program variable access (Sharp) [recommendation: to close] Arturo: it is covered by 890. - references to signals in a program which (16.2 - 890) Move: Gord - close mantis 564 (covered by 890) Second: Steven Abstain: none Opposed: none passed closed covered by 890 551 - Program block interaction with queues (Sharp) Gord - all except for possibly the DPI comment are addressed by 890. Steven - requesting to leave it open for now. Move: Gord: if there are no concerns from Cadence Steven can close 551 covered by 890 Second: Steven Abstain: none Opposed: none passed AI:Steven - provide feedback on 551 by next meeting. 1717 - 15.14.2 Clocking block driving value resolution - request for change (Jonathan) Move: Jonathan: close mantis 1717 (completely covered by 890) Second: Heath Abstain: none Opposed: none passed closed covered.to 890 1715 - Triggered property of a clocking block (Arturo - Jonathan) Neil: it uses a phrase "a simple value", needs to be changed (not LRM-like language) Arturo - triggered property --> triggered method Gord - happens later, region-wise compared to the event Neil - backward compatibility problem can't have a clockvar named triggered Jonathan - would be better to use an event handle (Arturo had told him there wouldn't be an easy way to do that). Dave - doesn't see a compelling need for adding this. Gord - makes it symmetric with other events. Jonathan - his goal is to get the same effect of ##0 within a class using a virtual interface. Dave - why not use repeat (@cb)? Jonathan - can't do ##0 that way; At the beginning of a BFM. Arturo - could refer to it as the event property. 14.5.4 Agreed to leave mantis 1715 open for now. 240 - Expression evaluation with cycle_delay (eugene) Jonathan - covered by the 'expect' construct - cute, but weird for verilog Arturo: the assertions covers this type of checking the context they are using it is procedural statement. combination of assertion and expect. Move: Arturo: close mantis 240 (covered by assertions and expect) Second: Jonathan Abstain: none Opposed: none passed closed 240. * 1325 - was closed by SVAC - unnamed clocking blocks [sv-ac] 1681 by svac: global clock Arturo: wrong solution to the real problem. looks like "timescale", quietly viral in nature Dave: looks more like a tool issue, not language issue Neil: motivation is to make it simpler to write assertions. Arturo - thinks it is more for a formal tool Gord: if for FV, should use attributes or something like that. - e.g. attributes on a cb Arturo - can we ask for more input on what the motivation is? Gord - if no default clocking, and using default, you would pick up the global... Would affect the universe of design. - if an attribute, fv would be able to handle designs that simulation can't, e.g. global clk only rec by the fv. This is probably a bad thing. Steven - we don't want to standardize simulation behavior on attributes Gord - there are things about global clocking that bother him. - how global? i.e. compilation unit versus design. Jonathan - thinks they meant simulation global. Gord - different global clock in different parts of system? sub-portion can have a global clock? Ray - thinks that type of requirement would follow. Jonathan - doesn't think it makes sense to have more than one global clk. Heath - could see using a global clock in different parts of sub-system. Jonathan - thinks that all clocks must align to the global clock. - this allows formal tools to handle multiple clock domains. Gord - fv restriction or a physical restriction? Arturo - only a formal verification issue. Gord - for simulation, not a problem to have unaligned clocks. - should tool restrictions be embedded in the LRM? - this type of stuff happens with synthesis all the time (what you can and can't do with synthesis) Neil: un-aligend clocks are reality. Jonahathn: 1530 has the copy of the slides for Dimitry. AI: Jonathan - send list of issues from svec to svac. 5. Additional mantis items ------------------------------------- The following items have proposals. 1371 Semantic of program block $exit (DaveR) -- Clarification Neil: the words 'can' should be changed to 'may' - cannot --> shall not - Verilog 2005 uses the word "may", v2k uses the word "can" Steven: the should be deleted - last sentence of first set of changes. - needs to be clarified. - current wording is misleading Francoise: you can not call $exit from a module. it has no effect. Gord - "the simulation shall terminate in this manner." Neil - reorder the sentence - start with "in order to..." Dave - "For the simulation to terminate in this manner...." Steven - "Simulations shall terminate in this manner... Jonathan - the problem is that we don't have words to distinguish between the obligations of the user versus the tool. Steven - the rewording is an improvement. - new issue - disable fork - only covers descendents, but not initial blocks. - Does killing an initial block imply that program blocks have reached their end? Steven - disable name is different than disable fork. - initial blocks have no scope, unless they have a begin end (and some variables?). - Should say "disable fork terminates initial processes" Dave - shall issue a disable fork and terminate all initial blocks in that program block. Steven - not all terminates happen because we have reached the end. Francoise: $exit not always a $finish Steven - only if last program with live initial blocks. AI: Dave - update 1371. AI:Steven - re-read in context of that section to see if any more issues AI: - ask Stu about (may versus can)? 6. Next meeting -------------------------------------- Monday March 19 2007 11:00-1:00pm ========== ADDITION TO the minutes, not part of regular ========= ========== meeting discussions, informational purposes ========= ========== Action Items updated based on March 5 2007 ==== Action items: March 5 2007 ------------------------- LRM and mantis item review, P1800 related items AI: Neil: send email on 890 to ac,bc,cc (post Cliff's portion on mantis) AI: Neil: forward Dmitry's slides to the active set of people on svec. (get them from Dennis) 890 and related items: AI: Neil - send feedback to Doug on any remaining outstanding issues with 890. - ask for feedback by Wed of next week from other committees. - add Cliff's writeup to mantis 890 (part 2) - have other committees review what is already on mantis from Doug AI: Mantis 551: Steven: Provide feedback on 551 AI: Mantis 1681 [svac] Jonathan -send list of issues from svec to svac. AI: Mantis 1371 Dave: update 1371 proposal Steven: re-read in context of that section to see if any more issues All: ask Stu about (may versus can)?