RE: Jumpered ports


Subject: RE: Jumpered ports
From: Kevin Cameron x3251 (dkc@galaxy.nsc.com)
Date: Fri Mar 15 2002 - 16:14:07 PST


>
> Thanks Dave,
> This is an excellent capture of the meeting. After you left, the following happened and should be reflected in the minutes:
>
....
>
> 2- Stefen has created a new issue associated with inout jumper port. Deadline is March 25 for discussion and review.
>
> Best Regards
> Vassilios

I'm currently trying to get some discussion going on this in the AMS committee as the
inabilty to express that two logical wires are the same physical wire is potentially
nasty. Digital verilog users usually use an "assign" to do bus-ripping e.g.:

  assign a = b[0]; // rip LSB

- which works fine because signal flow is usually unidirectional, but isn't an
explicit short-circuit, so in an AMS simulation 'a' and 'b[0]' could become two
nodes in the analog simulator with an A2D and D2A to evaluate the "assign" - which
is both inaccurate and inefficient if a short-circuit is intended.

Since the only mechanism for creating short-circuits appears to be jumpered ports,
and that appears to be getting ugly syntactically I would like to suggest that
we add an explicit short-circuit statement e.g.:

  alias a,b[0];

BNF ~:

  <alias_statement> ::== alias <net_lvalue>{,<net_lvalue>}

I chose "alias" because "join","connect" and "short" are all used for something
already.

That would remove the requirement for supporting jumpers in the port syntax (which
should be depracated in my opinion).

A short-circuit statement may also be useful if it can be used in a "generate"
statement, e.g. you could declare an unconnected n-dimensional array of instances
and then wire them seperately with multiple generate/alias statements.

Kev.



This archive was generated by hypermail 2b28 : Fri Mar 15 2002 - 16:16:05 PST