RE: [sv-bc] Ballot for proposed changes for 1800-2008 Draft 3

From: Stuart Sutherland <stuart_at_.....>
Date: Thu Apr 26 2007 - 10:37:51 PDT
There are several simulator and synthesis tools that use "memory" or
"memories" in tool-generated messages involving one-dimensional arrays.  If
the term is removed from the LRM and not from software tools, user's will
have no way to find out what these messages are referring to.  I would be
much more inclined to support removing the definition of what a "memory" is
from the LRM if I thought that at least the EDA vendors who are active in
the SV standards work would actually support removing the term from their
tools.  Maybe I'm just being pessimistic, but I don't think any vendor will.

I would also support the CC committee deprecating VPI memory access, or at
least adding it to the annex on items under consideration for deprecation.

Stu
~~~~~~~~~~~~~~~~~~~~~~~~~
Stuart Sutherland
Sutherland HDL, Inc.
stuart@sutherland-hdl.com
503-692-0898
 

> -----Original Message-----
> From: owner-sv-bc@server.eda.org 
> [mailto:owner-sv-bc@server.eda.org] On Behalf Of Bresticker, Shalom
> Sent: Thursday, April 26, 2007 1:36 AM
> To: Warmke, Doug; Brad Pierce; sv-bc@server.eda.org
> Subject: RE: [sv-bc] Ballot for proposed changes for 1800-2008 Draft 3
> 
> I think it remained in 1364-2001/2005 only because of PLI 
> back-compatibility. 
> 
> We could have handled other references to memories, e.g., 
> $readmem, without much trouble.
> 
> It needs to be SV-CC's decision.
> 
>  
> 
> Shalom
> 
>  
> 
> ________________________________
> 
> From: Warmke, Doug [mailto:doug_warmke@mentor.com] 
> Sent: Thursday, April 26, 2007 11:04 AM
> To: Bresticker, Shalom; Brad Pierce; sv-bc@server.eda.org
> Subject: RE: [sv-bc] Ballot for proposed changes for 1800-2008 Draft 3
> 
>  
> 
> Hello all,
> 
>  
> 
> Personally I would like to take advantage of the merge to remove the
> 
> concept of "memory" from the LRM.  It is a perfect subset of unpacked
> 
> fixed-size arrays.   Such redundancy can only lead to confusion and
> 
> reduce the LRM's integrity.
> 
>  
> 
> There are implications in PLI, as well as other places in the LRM.
> 
> Currently the SV-CC is working through some difficult issues
> 
> with arrays. I think the idea of removing "memory" as a special kind
> 
> of array wouldn't be too much additional work, and may even help
> 
> them simplify their problem space.
> 
>  
> 
> I think we could make this happen.
> 
>  
> 
> Regards,
> 
> Doug
> 
>  
> 
>  
> 
>  
> 
> ________________________________
> 
> From: owner-sv-bc@server.eda.org on behalf of Bresticker, Shalom
> Sent: Wed 4/25/2007 11:54 PM
> To: Brad Pierce; sv-bc@server.eda.org
> Subject: RE: [sv-bc] Ballot for proposed changes for 1800-2008 Draft 3
> 
> 1364-2005 has the following:
> 
> 4.9.3 Memories
> 
> A one-dimensional array with elements of type reg is also called a
> memory. These memories can be used to model read-only memories (ROMs),
> random access memories (RAMs), and reg files. Each reg in the array
> is known as an element or word and is addressed by a single 
> array index.
> 
> An n-bit reg can be assigned a value in a single assignment, but a
> complete memory cannot. To assign a value to a memory word, an index
> shall be specified. The index can be an expression. This 
> option provides
> a mechanism to reference different memory words, depending on 
> the value
> of other variables and nets in the circuit. For example, a program
> counter reg could be used to index into a RAM.
> 
> 4.9.3.1 Array examples
> 
> 4.9.3.1.1 Array declarations
> 
> reg [7:0] mema[0:255]; // declares a memory mema of 256 8-bit
>                        // registers. The indices are 0 to 255
> 
> Shalom
> 
> 
> > -----Original Message-----
> > From: owner-sv-bc@server.eda.org [mailto:owner-sv-bc@server.eda.org]
> > On Behalf Of Brad Pierce
> > Sent: Thursday, April 26, 2007 9:50 AM
> > To: sv-bc@server.eda.org
> > Subject: Re: [sv-bc] Ballot for proposed changes for 
> 1800-2008 Draft 3
> >
> > 7.2.4 in the draft under review begins
> >
> >    "A one-dimensional array with elements of types reg, logic or bit
> > is
> > also called a memory."
> >
> > In SystemVerilog this would include
> >
> >     logic M[1024];
> >
> > but not
> >
> >     reg [15:0] M[1024];
> >
> > -- Brad
> >
> > -----Original Message-----
> > From: Bresticker, Shalom [mailto:shalom.bresticker@intel.com]
> > Sent: Wednesday, April 25, 2007 11:38 PM
> > To: Brad Pierce; sv-bc@eda.org
> > Subject: RE: [sv-bc] Ballot for proposed changes for 
> 1800-2008 Draft 3
> >
> > Stu meant there is one unpacked dimension.
> > He did not say there are no packed dimensions.
> >
> > So this:
> >
> > >       reg [15:0] M[1024];
> >
> > is and always has been a memory.
> >
> > I don't know about this:
> >
> > > Is the following a memory?
> > >
> > >       reg [3:0] [3:0] M[1024];
> >
> > Verilog never had anything like this, so memories never 
> related to it.
> >
> > Shalom
> >
> > > -----Original Message-----
> > > From: owner-sv-bc@server.eda.org 
> [mailto:owner-sv-bc@server.eda.org]
> > > On Behalf Of Brad Pierce
> > > Sent: Thursday, April 26, 2007 9:26 AM
> > > To: sv-bc@server.eda.org
> > > Subject: RE: [sv-bc] Ballot for proposed changes for 
> 1800-2008 Draft
> > 3
> > >
> > > Stu writes --
> > >
> > > >> BP1-7-1   yes ___ no _X_ abstain ___
> > > >I disagree with the suggested change.  This subclause is
> > specifically
> >
> > > >defining a "memory" as a one-dimensional unpacked array. 
>  There is
> > no
> >
> > > >"fastest varying unpacked dimension".  An array with multiple
> > > unpacked
> > > >dimensions is not a memory, and referring to one as such would
> > break
> > > the VPI
> > > >definitions of memories and arrays.
> > >
> > > Yet the sentence in question is --
> > >
> > >   "Each packed dimension in the array is known as a memory element
> > or
> > > word"
> > >
> > > If there's only one dimension, then why is it talking about 'each
> > > packed dimension'?  And how could a dimension, let alone 
> each packed
> > > dimension, be an element?  And why isn't the following a memory?
> > >
> > >       reg [15:0] M[1024];
> > >
> > > It is an array with one unpacked dimension, but it is a two-
> > > dimensional array.
> > >
> > > Is the following a memory?
> > >
> > >       reg [3:0] [3:0] M[1024];
> > >
> > > -- Brad
> > >
> > > -----Original Message-----
> > > From: owner-sv-bc@eda.org [mailto:owner-sv-bc@eda.org] On 
> Behalf Of
> > > Stuart Sutherland
> > > Sent: Tuesday, April 24, 2007 11:47 PM
> > > To: sv-bc@eda.org
> > > Subject: RE: [sv-bc] Ballot for proposed changes for 
> 1800-2008 Draft
> > 3
> > >
> > > Matt,
> > >
> > > Here's my votes.
> > >
> > > I am traveling the morning of the conference call to discuss the
> > > voting.
> > > I should be able to call in for the first 30 minutes of the call.
> > >
> > > Stu
> > > ~~~~~~~~~~~~~~~~~~~~~~~~~
> > > Stuart Sutherland
> > > Sutherland HDL, Inc.
> > > stuart@sutherland-hdl.com
> > > 503-692-0898
> > >
> > >
> > > > SB1-1-1   yes _X_ no ___ abstain ___
> > > > SB1-1-2   yes _X_ no ___ abstain ___
> > > > SB1-1-3   yes _X_ no ___ abstain ___
> > > > SB1-1-4   yes _X_ no ___ abstain ___
> > > > SB1-1-5   yes _X_ no ___ abstain ___
> > > > SB1-2-1   yes ___ no _X_ abstain ___
> > > There are many references in the LRM to "1364" and "1800" 
> without a
> > > specific year.  Either these references need to be changed to
> > > "1364-2005" or "1800-2005", or the definition of "IEEE 1364" and
> > "IEEE
> >
> > > 1800" should be left in.  Most references to "1364" and "1800" can
> > > have a year added, but some may need to stay generic; specifically
> > > 1.7, 5.5.2, 21.12 (multiple places), C.2.1, C.2.2, I.12, I.12.1
> > > (multiple places).  I will change my vote to yes if the committee
> > > decides that all references to "1364" and "1800" have a specific
> > > version added.
> > >
> > > > SB1-3-1   yes ___ no _X_ abstain ___
> > > I have no problems with the current wording, and no new wording is
> > > recommended.  I will change my vote to yes of specific wording
> > changes
> >
> > > are recommended.
> > >
> > > > SB1-3-2   yes _X_ no ___ abstain ___
> > > > SB1-3-3   yes _X_ no ___ abstain ___
> > > A question is raised as to whether "config" should be listed as a
> > > design unit, but no recommendation given.  I will change 
> my vote to
> > > yes if the committee agrees to remove "config", or for a 
> description
> > > of config to be added to clause 3 (saying editor is to create the
> > > wording is OK).
> > >
> > > > SB1-5-1   yes _X_ no ___ abstain ___
> > > > SB1-5-2   yes _X_ no ___ abstain ___
> > > > SB1-5-3   yes ___ no _X_ abstain ___
> > > I think the discussion of time values makes more sense where it is
> > > than in clause 4.
> > >
> > > > SB1-5-4   yes _X_ no ___ abstain ___
> > > > SB1-5-5   yes _X_ no ___ abstain ___
> > > > SB1-5-6   yes _X_ no ___ abstain ___
> > > > SB1-6-1   yes _X_ no ___ abstain ___
> > > > SB1-6-2   yes _X_ no ___ abstain ___
> > > > SB1-6-3   yes _X_ no ___ abstain ___
> > > The suggestion is: "6.14 and 7.6 both describe strings. Merge them
> > > ***or*** have only a two-sentence paragraph in 6.14".  I do not
> > > approve of removing 6.14.  I do approve of I approve of having a
> > > two-sentence paragraph in 6.14 with a forward reference to 7.6.
> > >
> > > > SB1-7-1   yes _X_ no ___ abstain ___
> > > > SB1-7-2   yes _X_ no ___ abstain ___
> > > > SB1-10-1  yes ___ no _X_ abstain ___
> > > This needs to be a separate mantis item that goes through the full
> > > discussion, voting, and approval process.
> > >
> > > > SB1-10-2  yes _X_ no ___ abstain ___
> > > > SB1-11-1  yes ___ no _X_ abstain ___
> > > I think a more correct fix is to change the term "procedural
> > > assignment operators" in 10.3 and 10.3.1 to "assignment 
> operators",
> > so
> >
> > > as to match the description of these operators in 11.2.7 
> and A.6.3.
> > >
> > > > SB1-11-2  yes ___ no _X_ abstain ___
> > > The description of the wording problem is not detailed enough for
> > the
> > > editor
> > > (me) to implement.
> > >
> > > > SB1-11-3  yes ___ no _X_ abstain ___
> > > This is an open mantis item (#1004), and needs to be 
> addressed using
> > > the standard discussion, voting and approval method.
> > >
> > > > SB1-11-4  yes _X_ no ___ abstain ___
> > > > SB1-12-1  yes _X_ no ___ abstain ___
> > > > SB1-16-1  yes _X_ no ___ abstain ___
> > > > SB1-19-1  yes _X_ no ___ abstain ___
> > > > SB1-19-2  yes ___ no _X_ abstain ___
> > > This is an alleged errata in 1364-2005, and should be 
> addressed as a
> > > separate Mantis item.
> > >
> > > > SB1-20-1  yes _X_ no ___ abstain ___
> > > > SB1-20-2  yes _X_ no ___ abstain ___
> > > > SB1-21-1  yes ___ no _X_ abstain ___
> > > This is an enhancement request that needs to be a separate Mantis
> > > item.
> > >
> > > > SB1-21-2  yes ___ no _X_ abstain ___
> > > This is a change to a definition in 1364-2005.  It should be a
> > > separate Mantis item for discussion, voting and approval.
> > >
> > > > SB1-21-3  yes _X_ no ___ abstain ___
> > > > SB1-22-1  yes _X_ no ___ abstain ___
> > > > SB1-22-2  yes _X_ no ___ abstain ___
> > > > SB1-22-3  yes _X_ no ___ abstain ___
> > > > SB1-22-4  yes ___ no _X_ abstain ___
> > > I agree with the principle of the change, but this is a definition
> > > that does not exist in either 1364-2005 or 1800-2005.  It needs to
> > be
> > > a separate Mantis item with discussion, voting and approval.
> > >
> > > > SB1-25-1  yes _X_ no ___ abstain ___
> > > > SB1-25-2  yes _X_ no ___ abstain ___
> > > > SB1-25-3  yes _X_ no ___ abstain ___
> > > > SB1-27-1  yes _X_ no ___ abstain ___
> > > > SB1-33-1  yes ___ no ___ abstain _X_
> > > I think the current order of clauses 30-33 is fine, but 
> will change
> > > order if the committee approves.
> > >
> > > > SB-O-1    yes ___ no _X_ abstain ___
> > > This needs committee discussion on what the correction should be.
> > > Make
> > > it a
> > > separate Mantis item.
> > >
> > > > SB-O-2    yes ___ no _X_ abstain ___
> > > This needs committee discussion on what the correction should be.
> > > Make
> > > it a
> > > separate Mantis item.
> > >
> > > > SB-O-3    yes ___ no _X_ abstain ___
> > > This needs committee discussion on what the correction should be.
> > > Make
> > > it a
> > > separate Mantis item.
> > >
> > > > SB-O-4    yes ___ no _X_ abstain ___
> > > I think the current split on string literal topics is 
> correct.  5.9
> > > discusses string literal syntax in the same clause with other
> > > literals.
> > > 11.6.1 discusses operations on string literals in the 
> same clause on
> > > operations on other types of data.
> > >
> > > > SB-O-5    yes ___ no _X_ abstain ___
> > > Moving the discussion on string literals and string types closer
> > > together because they are related would only move other closely
> > > related topics further apart.  This is one of those cases where
> > there
> > > is no perfect solution to how to group topics.  I prefer to keep
> > > string literals with other literals and string arrays with other
> > > arrays.
> > >
> > > > SB-O-6    yes ___ no _X_ abstain ___
> > > This needs committee discussion on what the correction should be.
> > > Make
> > > it a
> > > separate Mantis item.
> > >
> > > > SB-O-7    yes ___ no _X_ abstain ___
> > > This needs committee discussion on what the correction should be.
> > > Make
> > > it a
> > > separate Mantis item.
> > >
> > > > SB-O-8    yes _X_ no _X_ abstain ___
> > > This is a duplicate of SB1-10-2, of which I am also in favor.
> > >
> > > > SB-O-9    yes ___ no _X_ abstain ___
> > > This needs committee discussion on what the correction should be.
> > > Make
> > > it a
> > > separate Mantis item.
> > >
> > > > SB-O-10   yes ___ no _X_ abstain ___
> > > I think this should be left as-is.
> > >
> > > > SB-O-11   yes ___ no _X_ abstain ___
> > > Discussion on this should be turned over to the SV-CC committee.
> > >
> > > > SB2-6-9-1 yes _X_ no ___ abstain ___
> > >
> > >
> > > > BP1-5-1   yes ___ no _X_ abstain ___
> > > The note in 5.8 should remain deleted.  It is redundant with the
> > same
> > > note in 5.7.  The Greek letter has already been corrected 
> for draft
> > 3.
> > >
> > > > BP1-5-2   yes _X_ no ___ abstain ___
> > > > BP1-5-3   yes ___ no _X_ abstain ___
> > > This would be changing a definition in 1800-2005.  It is also part
> > of
> > > a larger issue that has been raised on consistent usage of "data
> > type"
> > > and
> > > "data object" throughout the LRM, and should be dealt with in a
> > Mantis
> >
> > > item on that topic.
> > >
> > > > BP1-5-4   yes ___ no _X_ abstain ___
> > > This is part of a larger issue that has been raised on consistent
> > > usage of "data type" and "data object" throughout the LRM, and
> > should
> > > be dealt with in a Mantis item on that topic.
> > >
> > > > BP1-5-5   yes ___ no _X_ abstain ___
> > > This is part of a larger issue that has been raised on consistent
> > > usage of "data type" and "data object" throughout the LRM, and
> > should
> > > be dealt with in a Mantis item on that topic.
> > >
> > > > BP1-5-6   yes _X_ no ___ abstain ___
> > > > BP1-5-7   yes ___ no _X_ abstain ___
> > > This is part of a larger issue that has been raised on consistent
> > > usage of "data type" and "data object" throughout the LRM, and
> > should
> > > be dealt with in a Mantis item on that topic.
> > >
> > > > BP1-5-8   yes ___ no _X_ abstain ___
> > > This is part of a larger issue that has been raised on consistent
> > > usage of "data type" and "data object" throughout the LRM, and
> > should
> > > be dealt with in a Mantis item on that topic.
> > >
> > > > BP1-6-1   yes _X_ no ___ abstain ___
> > > > BP1-6-2   yes ___ no ___ abstain _X_
> > > I will go along with what the committee decides.
> > >
> > > > BP1-6-3   yes _X_ no ___ abstain ___
> > > > BP1-6-4   yes _X_ no ___ abstain ___
> > > > BP1-6-5   yes ___ no _X_ abstain ___
> > > This is already covered in Mantis 507, and is scheduled to be
> > > corrected in draft 3.
> > >
> > > > BP1-6-6   yes ___ no ___ abstain _X_
> > > I will go with what the committee decides regarding removal of the
> > > editor question, but personally I do not think this is an
> > appropriate
> > > usage of an informative note.  I think this note should either be
> > > deleted or be made part of the preceding paragraph.
> > >
> > > > BP1-7-1   yes ___ no _X_ abstain ___
> > > I disagree with the suggested change.  This subclause is
> > specifically
> > > defining a "memory" as a one-dimensional unpacked array.  There is
> > no
> > > "fastest varying unpacked dimension".  An array with multiple
> > unpacked
> >
> > > dimensions is not a memory, and referring to one as such 
> would break
> > > the VPI definitions of memories and arrays.
> > >
> > > > BP1-7-2   yes _X_ no ___ abstain ___
> > > > BP1-7-3   yes _X_ no ___ abstain ___
> > > > BP1-7-4   yes ___ no _X_ abstain ___
> > > I disagree with removing the paragraph.  An introduction paragraph
> > and
> >
> > > general definition of a structure is needed.  If the text is not
> > > correct, then a proposal on correct wording should be made,
> > discussed,
> >
> > > and voted on.
> > >
> > > > BP1-8-1   yes _X_ no ___ abstain ___
> > > > BP1-9-1   yes _X_ no ___ abstain ___
> > > > BP1-9-2   yes _X_ no ___ abstain ___
> > > > BP1-9-3   yes _X_ no ___ abstain ___
> > > > BP1-9-4   yes _X_ no ___ abstain ___
> > > > BP1-10-1  yes _X_ no ___ abstain ___
> > > > BP1-10-2  yes _X_ no ___ abstain ___
> > > > BP1-10-3  yes _X_ no ___ abstain ___
> > > > BP1-11-1  yes _X_ no ___ abstain ___
> > > > BP1-12-1  yes ___ no ___ abstain _X_
> > > I will go along with the committee decision.  I think it 
> is helpful,
> > > but not essential, to mention that a default branch is 
> not the only
> > > way to catch x/z values in the case expression.
> > >
> > > > BP1-12-2  yes _X_ no ___ abstain ___
> > > > BP1-12-3  yes _X_ no ___ abstain ___
> > > > BP1-12-4  yes _X_ no ___ abstain ___
> > > > BP1-12-5  yes _X_ no ___ abstain ___
> > > > BP1-13-1  yes _X_ no ___ abstain ___
> > > > BP1-13-2  yes _X_ no ___ abstain ___
> > > > BP1-13-3  yes _X_ no ___ abstain ___
> > > > BP1-13-4  yes _X_ no ___ abstain ___
> > > > BP1-13-5  yes _X_ no ___ abstain ___
> > > > BP1-13-6  yes _X_ no ___ abstain ___
> > > > BP1-13-7  yes _X_ no ___ abstain ___
> > > > BP1-22-1  yes _X_ no ___ abstain ___
> > > > BP1-22-2  yes _X_ no ___ abstain ___
> > > > BP1-22-3  yes _X_ no ___ abstain ___
> > > > BP1-22-4  yes _X_ no ___ abstain ___
> > > > BP1-22-5  yes _X_ no ___ abstain ___
> > > > BP1-A-1   yes _X_ no ___ abstain ___
> > > > MH-1      yes ___ no _X_ abstain ___
> > > This is a duplicate of BP1-7-4.  A Mantis item should be created
> > with
> > > a proposal for the exact changes needed.
> > >
> > > > MH-2      yes ___ no _X_ abstain ___
> > > A Mantis item should be created with a proposal for the exact
> > changes
> > > needed (one Mantis item could cover both MH-1 and MH-2).
> > >
> > > > MH-3      yes _X_ no ___ abstain ___
> > > I approve based on the assumption that the requested change is to
> > > delete the words "Verilog syntax for" instead of the currently
> > deleted
> >
> > > "Verilog syntax".
> > >
> > > > HC-D-1    yes _X_ no ___ abstain ___
> > > > HC-D-2    yes _X_ no ___ abstain ___
> > > > HC-E-1    yes _X_ no ___ abstain ___
> > > > HC-O-1    yes _X_ no ___ abstain ___
> > > > HC-O-2    yes _X_ no ___ abstain ___
> > > > HC-O-3    yes _X_ no ___ abstain ___
> > > > HC-O-4    yes _X_ no ___ abstain ___
> > > > HC-O-5    yes _X_ no ___ abstain ___
> > > > HC-O-6    yes _X_ no ___ abstain ___
> > > > HC-O-7    yes _X_ no ___ abstain ___
> > >
> > > Most of Heath's issues were about formatting.  The note at the
> > > beginning of these annexes already states that formatting still
> > needs
> > > to be done.
> > > No
> > > review or voting on this type of issue was needed.
> > >
> > >
> > >
> > >
> > >
> > > --
> > > This message has been scanned for viruses and dangerous content by
> > > MailScanner, and is believed to be clean.
> > >
> > >
> > > --
> > > This message has been scanned for viruses and dangerous content by
> > > MailScanner, and is believed to be clean.
> >
> > --
> > This message has been scanned for viruses and
> > dangerous content by MailScanner, and is
> > believed to be clean.
> 
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
> 
> 
> 
> 
> -- 
> This message has been scanned for viruses and 
> dangerous content by MailScanner 
> <http://www.mailscanner.info/> , and is 
> believed to be clean. 
> 



-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Thu Apr 26 10:38:27 2007

This archive was generated by hypermail 2.1.8 : Thu Apr 26 2007 - 10:38:56 PDT