Re: IBIS IRD68 - Clarify that Rising and Falling Waveforms Should be Correlated

On Fri, 03 Nov 2000, Matthew Flora wrote:
> I'm concerned about using older models where the [Rising Waveform]'s within a
> [Model] are time correlated to each other and the [Falling Waveform]'s within
> that [Model] are time correlated to each other, but no time correlation exists
> between the [Rising waveform]'s and the [Falling Waveform]'s. How is the
> simulator to recognize one of these models made prior to the clarification and
> a model made after the clarification? Would it be based upon the [IBIS_ver]
> value?

It doesn't matter.

With IBIS as it has been, this correlation is undefined. A simulator
must create its own correlation, because without it it would be
impossible to support diffrerential drivers or driver schedule, and
and it would be difficult to do periodic signals. There is no basis
to say that any other method might be superior to this one.

Therefore, some of the existing simulators already do it this way, so
are already in conformance with no changes. Others that used a
different method of establishing correlation would have a choice.
They could key off [IBIS_ver], or they could apply the Bird68 method
everywhere. Simulations will be more consistent vendor to vendor if
the Bird68 method is applied everywhere. For simulators that did it
a different way, the method required here is usually a
simplification. Models that assumed a different correlation have
always simulated different on different simulators, with some
matching what the creator hoped for, and some not.

This Bird does not cause anything to break that wasn't broken
already, and it makes an important clarification. I am guessing
that the reason for proposing this is the author's frustration with
the arbitrary differences caused by the fact that this is not defined.
It is frustrating for simulator developers and modelers, too. Bird68
should be approved.

Open question for simulator vendors:

1. Is yours already in conformance?

2. If not, how do you do it?

As a member of the IBIS futures subcommittee, I would appreciate
knowing this. You can reply privately if you don't want the world to
know.

 
Received on Fri Nov 3 17:20:04 2000