Re: [sv-bc] RE: Email Vote: respond by 8AM PDT, Friday, July 30, 2010

Hi Shalom.

For VCS, at least, the choice of min vs. typ vs. max can be made
by a command-line switch. I tried reinvoking VCS with +maxdelays
on a small design with a timing triplet in a parameter, and it
did recompile, with the message,

  "recompiling module junk because:
        Module parameters have been changed via defparam."

It probably is an elaboration choice for VCS, as you say.

On 08/03/2010 11:48 PM, Bresticker, Shalom wrote:
> Hi, John.
>
> I don't think the LRM says, at least explicitly, when the min-typ-max choice is made.
> In practice, in the tools I see, the choice is made at elaboration time, not at run time.
> In fact, the choice has to be so, because mintypmax expressions can appear in constant expressions that the elaboration is dependent on.
>
> In any case, the form
>
> parameter P = 1:2:3;
>
> is syntactically and semantically legal and has been so for many years.
>
> What the 2005 LRM said explicitly and got omitted from the 2009 standard was that if the RHS result value is not a real or an integral value (e.g., string, array), then the parameter declaration has to explicitly include the data type of the parameter.
>
> Regards,
> Shalom
>
>> -----Original Message-----
>> From: owner-sv-bc@eda.org [mailto:owner-sv-bc@eda.org] On Behalf Of
>> John Michael Williams
>> Sent: Wednesday, August 04, 2010 4:59 AM
>> Cc: sv-bc@eda.org
>> Subject: Re: [sv-bc] RE: Email Vote: respond by 8AM PDT, Friday, July
>> 30, 2010
>>
>> Hi Steve.
>>
>> I agree that a min:typ:max is an expression (of an alternation set
>> of numerical values).
>>
>> The min vs typ vs max choice is taken at runtime and is
>> dependent on the simulator. We are talking about
>> syntax here (at least, that was my intention).
>>
>> The compiled and elaborated module will represent a parameter with
>> a value of "x:y:x", in which x, y, and z will be numbers. The
>> implementation of x:y:z COULD be as you say; or it could be something
>> else, such as a string, linked list, or array.
>>
>> I agree that x:y:z could be a string value, or a subset value, and
>> would
>> NOT be a numerical value.
>>
>> Nevertheless, the assignment statement, parameter = x:y:z; should
>> be syntactically correct. That was what I meant to be the point
>> of my original Email.
>>
>> On 08/02/2010 07:09 PM, Steven Sharp wrote:
>>>
>>>> Date: Mon, 02 Aug 2010 10:28:55 -0700
>>>> From: John Michael Williams<john@svtii.com>
>>>
>>>> The new statement makes sense, because a parameter should
>>>> be able to store things such as delay triplets:
>>>>
>>>> parameter MyDelay = 3:4:5;
>>>
>>> A mintypmax_expression is not a value, nor would a parameter ever
>> store it.
>>> It is an expression, which evaluates to a single numerical value
>> based on
>>> which delay is selected during elaboration. The parameter would hold
>> the
>>> single value that was selected, not the triplet. It is similar to
>> writing
>>>
>>> parameter MyDelay = min_selected ? 3 : (typ_selected ? 4 : 5);
>>>
>>> The parameter doesn't hold that expression; it holds the result of
>> it.
>
> ---------------------------------------------------------------------
> Intel Israel (74) Limited
>
> This e-mail and any attachments may contain confidential material for
> the sole use of the intended recipient(s). Any review or distribution
> by others is strictly prohibited. If you are not the intended
> recipient, please contact the sender and delete all copies.
>
>

-- 
      John Michael Williams
      Senior Adjunct Faculty
Silicon Valley Technical Institute
-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.