SV-EC Committee Meeting Monday July 31 2006 11:00am - 1:00pm PST [Minutes distributed for approval at next (sv-ec) committee meeting] (1021121020213 ) Day (9933370825601 ) (1000000000000 ) Month (2112334556677 ) (0000000000000 ) Year (5666666666666 ) --------- Attendees ---------- (AAAAAAAAAAAAA ) Arturo Salz (-AAA-A-AAAAAA ) Cliff Cummings (A-AAAAAAAAAAA ) Dave Rich (AAAAAAAAAAAAA ) Francoise Martinolle (AAAAAAAAAAAAA ) Mehdi Mohtashemi (AAAAAAAAAAAAA ) Neil Korpusik (AAAAAAaAAAAAA ) Ray Ryan (-AAAAA-AAAA-A ) Gordon Vreugdenhil (----AAAAAaAAA ) Steven Sharp (-A--AA----A-- ) Phil Moorby (---AAA---A-A- ) Stu Sutherland (--AAAaaA----- ) Surrendra Dudani (AA----------- ) Doug Warmke (-----a------- ) Don Mills (Aa----------- ) Chris Spear (-----------AA ) Heath Chambers ^^------------ Considered ONE MEETING. ** Minutes taken by Neil Korpusik and Mehdi Mohtashemi ////////////////// July 31, 2006 ///////////////////////// Logistics note: You can use *6 for mute/unmute key sequence Agenda: 1. IEEE patent policy ref: http://standards.ieee.org/board/pat/pat-slideset.ppt Move: Cliff- Assume that the patent policy was read Second: Neil Abstain: none Opposed: none passed 2. Review Meeting minutes, of and July 10th 2006 http://www.eda.org/sv-ec/Minutes/SV-EC_Meeting_July_10_2006_Minutes.txt Move: Dave- Approve the minutes of July 10th, 2006 Second: Neil Abstain: Gord (was not at July 10th meeting) Opposed: none passed 3. Action items review Action items: July 10, 2006 ------------------------- AI: 978 Neil - Take sentence out of 5.15.1, add to 5.15. Take sentence out of 5.15.3 proposal. <--- this part done? [DONE] AI: 1517 Dave: mantis item for partitioning constraints for array locator [DONE] AI: 976 Dave - change section 5.9.4 to 5.9.5 in the proposal - say real or shortreal in change for 5.9.7 - show strike-outs for 5.10.7 and then repeat the modified text, without strike-outs for 5.10.8. methods. AI: 890 Cliff and all - send email to reflector to get the discussion started Cliff is working on it. Would like to have Doug present in the meeting. [DONE] AI: 1457 Dave make the modifications."treated as unsigned" and fix the typo. [DONE] AI: 1474 Mehdi - create a pdf for the proposal. Not everyone can see all of the changes From Previous meetings: AI: 1330 Gord - new language proposal to handle the port connection issues. AI: 1330 Dave - examples via email on the issue with ports in interfaces AI: 889 Arturo - create a mantis item describing how to handle the null handles for class, to be included in 13.3 Next to last sentence can go away -- blue in proposal. AI: 976 Dave - put a proper proposal in place [DONE] AI: 960 Mehdi - discuss this with Matt to resolve the overlap. AI: 1330 Dave, add the example list to do. AI: 261/262 combined Steven put together some examples of overflow -- Arturo unsigned/signed examples. AI: new item - Gordon to create a new mantis item for randomize() method call item (Ray, Gordon have discussed - still working on it internally) 4. Continue: Review mantis items with proposals -------------------------------------------------- 978 5.15 array methods, minor errata (Shalom/Neil) Neil - Made several updates in version 5 of the proposal. - There are several other mantis items that touch the same 3-pages of the LRM. 1038, 1437, 1490, 1517, 1457 - 1038 - multi-dimensional issues - Ralph Duncan. - covered by 1490 and 978 CLOSE it next meeting Mehdi - combine the following with 978? - 1457 - wildcards - has more than just 15.15 changes - we agreed not to combine it with 978 - 1490 - discussed below - can be merged with 978 section 5.15.2 1490 Array ordering methods on associative arrays (Dave) Discussion about Multi-dimensional arrays and how to word the ONE-Dimension. Neil - packed versus unpacked. Should we only allow unpacked for array ordering methods, as shown in the proposal? Explainatory NOTE: At first we were thinking that we should also allow packed, but after a fairly long discussion we decided to limit the array ordering methods to just unpacked arrays. The example that Gord gave was a key consideration for this decision. Gord: consistenty and which way for packed and unpacked - prefers just unpacked arrays for all of the array ordering methods - both of the following should be consistent. Prefers to not allow these methods on packed arrays as a means of making both of the following consistent. i.reverse() // integer i; i.reverse() // reg [31:0] i; packed array Arturo: favor consistency, discussion of how deep the method goes, currently ambiguity that which intent was to handle all dimensions, returning index, it is a single index. Dave - can use streaming to do a bit reversal of a packed array. reverse() appears to be useful for packed arrays. - need to be on the least varying dimension (leftmost dimension) Steven - arrays of arrays (just reordering that array). - LRM uses the phrase "multi-dimensional arrays", which causes confusion. - operates on leftmost dimension if view as a multi-dimensional array operates on the array if view as an array of arrays Arturo - arrays of arrays seems to be a reasonable view. Gord - arrays of arrays simplifies the discussions. - more like structural nesting (versus multi-dimensions) Steven - top level dim is leftmost, otherwise you are indexing to an element Arturo - yes, ok to limit some to one set of arrays. - the original intent was to handle all dimensions. Steven - How sort a 2-D entity? - locators could go down multiple dim. Francoise - should be one dimension only Gord - array of int - can ask for elements with value 5. Need to do it for each row. In 5.15 - nothing seems to operate on more than one-dim, but only on one vector of items. - can use bitstream operator to reverse bits of a packed array. - with(item.sum()) could be used to do multi-dimensions for reduction operators Steven - could use a cast inside the with-clause to take care of overflow with(item + 90'b0) // 90'b0 will widen the value The width of the expression of the with-clause will determine the width of the value returned. - allowing packed arrays should be ok, since you only ever go down one dimension. Gord - enum is a place where "methods" operate on individual bits Ray: is there a way to revers, butterfly calculation, operator to be synthesizable and reversible. Ray - had reservations about not allowing array reduction methods on packed arrays. Could back out this restriction later if necessary. AI/Gord - send an example 2-dimension of array and how to do the reduction method to Neil. AI/Neil - add an example to clarify how to deal with dimensions. how to write a reduction operation on an array of array of int. - reverse() - take out (unpacked or packed) from current LRM text. - merge 1490 into 978 - add text and example for specifying width of result (e.g. using with clause) "Array ordering methods reorder the elements of any unpacked array (fixed or dynamically sized) except an associative array." 1038: close it off next time, covered by 978 1457 Problems with Associative arrays and wildcard index type (Dave) Don't combine this one with 978. 1437 Prototype syntax for array locator methods in 5.15.1 is not legal(daver) (similar to 0082) Neil - This one deals with the BNF in 5.15.1 A.2.6 a function returns data_type A.2.2.1 data_type doesn't include unpacked arrays Dave - get rid of [$] and add a note on the syntax - no special semantics implied here, the syntax is just not correct - create an identifier and define it Francsoise - should be ok if we define array_type - the prototype seems to be more confusing that helpful Steven - array_type is "type of item" Gord - change array_type ---> element_type_of_array (ie of the array) - note that this is not legal syntax in LRM Arturo - take out the prototype Dave - leave this item separately (don't merge with 978) - change prototype to prose. AI/Dave - update 1437 also include a proposal 976 5.9-5.10 clarifications (Shalom, DaveR) Dave: made all changes. Move: Dave - Approve the proposal for mantis 976 Second: Gord: does the check need to be done at runtime or elaboration time? Francoise: looks like a runtime check, provide a warning. Mehdi: needs to be discussed further. 5. Next meeting August 14th, 2006 11-1 PST The following people can't make it to the next meeting. Cliff, Arturo, Heath 6. Name resolution Gord - the bc is starting up a sub-group on this. An email will go to the sv-ec on this. There will be some impact to sv-ec. Need folks to participate in this sub-group. ========== ADDITION TO the minutes, not part of regular ========= ========== meeting discussions, informational purposes ========= ========== Action Items updated based on July 31st ========= Action items: July 31, 2006 ------------------------- AI: 1490 Gord - send an example 2-dimension of array and how to do the reduction method to Neil. Neil - add an example to clarify how to deal with dimensions. how to write a reduction operation on an array of array of int. - reverse() - take out (unpacked or packed) from current LRM text. - merge 1490 into 978 - add text and example for specifying width of result (e.g. using with clause) AI: 1437 Dave - update 1437, also include a proposal ==================================================================================