Re: Bit selects of specparams?

From: Chong Guan Tan (tan@surefirev.com)
Date: Tue Jul 20 1999 - 15:37:55 PDT

> X-Authentication-Warning: gw.boyd.com: majordomo set sender to
owner-btf@boyd.com using -f
> MIME-Version: 1.0
> From: Ted Elkind <elkind@cadence.com>
> To: Stuart Sutherland <stuart@sutherland.com>
> Cc: wadswort@poci.amis.com, elkind@sphinx.cadence.com, btf@boyd.com,
drew@surefirev.com, roberts@sphinx.cadence.com, chas@sphinx.cadence.com,
gsrao@sphinx.cadence.com, sjmeyer@pragmatic-c.com
> Subject: Bit selects of specparams?
> Date: Tue, 20 Jul 1999 17:41:11 -0400 (EDT)
>
> Hi Stu,
>
> In case no one replied yet, specparams cannot be accessed with bit or
> part selects. They can only provide time or delay values or mtm
> triples. However, they can always appear anywhere in a time or delay
> expression. They cannot appear outside that context. Hence, this is
> legal:
>
> specparam x = 5:6;7;
> ...
> #x Q = DAT;
>
> But this is not:
>
> specparam x = 5:6;7;
> reg y;
> reg [7:0] z;
> ...
> y = z[x];
>
> And neither is this:
>
> specparam [7:0] x = 5:6:7;
>
> Ted

Sorry for jumping into the specparam discussion without warning. My
name is Tan chong, besides myself, no one knows me in the world of
Verilog.

I like to give my input on what is happening with specparam. It is
nice to have someone spec out what is not "accepted", well,
VXL accepts this design :
module test ;

-----
specify
  specparam[0:2] A = 3:4:5 ;
endspecify

initial
  $display( "A= %d", A ) ;

wire W1 = A[1] ;
endmodule 

---

Where A has a range specification, and used in non-delay 
context.

Better yet, VXL support this design:

---
module top ;
wire WT = test.A ;
endmodule

module test ;

<p>specify
  specparam[0:2] A = 3:4:5 ;
endspecify

initial
  $display( "A= %d", A ) ;

wire W1 = A ;
endmodule
-----

<p>I will say the sensible things to do is to not allow the use
of specparam outside the specify block.  I heard that the IEEE 
standard is allowing this feature, which I believe is wrong. I realize 
that there are some cell libraries from a major vendor in US that 
use this "feature", but the sole purpose of a committe is to 
set up a sensible standard,  define the meaning of the standard,
and get rid of all confusion that may arise due to poor
specification and implementation in the past.

<p><p>NO on specparam abuses group,
tan

This archive was generated by hypermail 2.1.4 : Mon Jul 08 2002 - 12:53:29 PDT and
sponsored by Boyd Technology, Inc.