Subject: RE: Inferring FPGA Hardware Features?
From: Gilbert Nguyen (nguyen@ibiquity.com)
Date: Tue Jul 17 2001 - 12:22:12 PDT
I will put in my 2 pennies worth:
I can see attributes for || case statement or
priority encoding etc. BUT "optimize for speed"
or "use certain cell" seems to be more a synthesis
constraint and not attributes.
I think if you know for sure that you are calling
for that cell then you need to instatiate it & use
ifdef to switch between libraries.
Gilbert Nguyen
-----Original Message-----
From: Michael McNamara [mailto:mac@verisity.com]
Sent: Monday, July 16, 2001 5:41 PM
To: Shalom Bresticker
Cc: mac@verisity.com; vlog-synth@eda.org
Subject: Re: Inferring FPGA Hardware Features?
What I am describing has to do with reuse -- Imagine that at one time
I wanted a particular transformation, or attribute to be attached to a
particular construct. At some later time when I reuse that construct
in a different way, I do not want the particular attribute attached
anymore. I could certainly edit the text the second time; but that
becomes tedious and error prone.
It is the case that certain attributes will remain for all time true
about a particular item; i.e., a case statement is parallel.
Other attributes, such as 'optimize for speed', or 'use cell 56' may
only be true for one use of the construct.
What I am ruminating about is whether we need conditional attributes.
wire [31:0] a (* if (PentiumIII) maxdriver=1;
else if (PentiumIV) maxdriver=2;
else maxdriver=10; */ ;
I am not convinced that if we provided this, that as a pratical matter
users will use it correctly; however without a method to conditionally
turn on or off certain attributes, they become less useful than they
could be.
-mac
Shalom Bresticker writes:
> [1 <text/plain; us-ascii (7bit)>]
> Mac,
>
> You're correct, but I'm not sure that is a problem.
>
> You say, "hence I don't get what I want".
> Well, if that's the case, then if you're the one developing the
module, then don't put it in.
> If you're the one wanting to reuse it, then it won't fit your needs.
>
> On the other hand, if the alternative is having to ADD that attribute
to get what you want,
> then it's also not reusable without change.
>
> It simply becomes one of the characteristics of that module.
>
> Am I way off base here ?
>
> Shalom
>
>
> Michael McNamara wrote:
>
> > My point is extremely pragmatic. If I insert into a module a
pragma
> > that states something, like (* implement this in cell 14 *), then
> > when I instatiate this module more than once, I will have two
cells
> > that are trying to use cell 14.
> >
> > If I state in a module, (* optimize for time *), then _every_
> > instatiation of that module will be requesting timing
optimization;
> > including instantiations that are not timing critical; hence I
don't
> > get what I want.
>
> --
>
************************************************************************
**
> Shalom Bresticker
Shalom.Bresticker@motorola.com
> Motorola Semiconductor Israel, Ltd. Tel #: +972 9
9522268
> P.O.B. 2208, Herzlia 46120, ISRAEL Fax #: +972 9
9522890
>
************************************************************************
**
>
>
> [2 <text/html; us-ascii (7bit)>]
>
This archive was generated by hypermail 2b28 : Tue Jul 17 2001 - 12:33:32 PDT