Re: Verilog-AMS Committee Meeting Reminder - 25 Jan 2007

From: Marq Kole <marq.kole_at_.....>
Date: Thu Jan 25 2007 - 05:57:19 PST

Geoffrey,

Thanks for the feedback.

I'll update the notion on the attribute. In the same section option 3 would use the attribute only to suppress the warning about a race condition. I guess that would then count as "acceptable" use of an attribute as it does not change the simulation results.

The diode example was chosen as sufficiently compact to be readable to someone not well versed in compact device models. Adding a strobe in the initialization module would be appropriate - in the handling of multirate analysis it should be well defined what could be considered for the sensitivity list of an analog block. In that sense my guess was that the analysis of the noise analysis block would reveal that the only contributions made in this block are noise contributions, and that the calculated variable values are only used by the noise contributions, hence the whole block only needs to be evaluated in case of a noise analysis. Maybe this presumes too much about the possibilities in synthesizing a sensitivity list for analog blocks.

Would it be possible for you to create a more condensed example from the PSPNQS model that could be used as an illustration of multiple analog block usage? I'd be happy to include it in the document.

Cheers,
Marq


Marq Kole
Competence Leader Robust Design

Research
NXP Semiconductors
Tel: +31 40 27 49051, Fax: +31 40 27 44700, Mobile: +31 6 387 48 389
High Tech Campus 48 p.2.039, 5656 AE Eindhoven, The Netherlands
marq.kole@nxp.com, www.nxp.com

"Geoffrey.Coram" <Geoffrey.Coram@analog.com> Sent by: geoffrey.coram@analog.com 25-01-2007 14:35

To Marq Kole <marq.kole@nxp.com> cc verilog-ams <verilog-ams@eda-stds.org> Subject Re: Verilog-AMS Committee Meeting Reminder - 25 Jan 2007 Classification

Hi, Marq -

In 3.1.1, option 2 suggests use of an attribute for marking
race conditions.  In 1364, attributes are not supposed to
be used for anything that can change simulation results;
they are in some sense "hints" to be used to help the
simulator perform more efficiently.  I think your option 2
would violate this definition.

The diode example (Ex 3 in 2.3) is not a very good one for
justifying multiple analog blocks.  For example, the
initialization block would not satisfy Martin's need to
have a way to $strobe a warning about a parameter only
once -- unless you add some semantics about how to
determine "sensitivity lists" for the blocks.  In the
noise section, the sensitivity list would indicate that
the block needs to be computed; if the pow() function
causes a math error, is this reported even if not doing
a noise analysis?

I thought the more compelling example was the non-quasi-static
(NQS) part of the PSP mos model, where based on a parameter
(SWNQS) one could have 0 to 9 points in or the spline
collocation method (and, with use of idt(), each point might
require 2 internal nodes).

-Geoffrey


--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean. Received on Thu Jan 25 05:57:26 2007