Bob,
That’s an invitation for me to step up on my “favorite” soap-box, and vent a
little…
This is all coming from the age-old IBIS rigidity. We were already suffering
from that
before IBIS-AMI was invented. There was a time when several of us feared that
IBIS was dead. Every time a new buffer architecture was invented since the
90’s,
IBIS was struggling to keep up and implement support for it. The new keywords,
such as [Driver Schedule], [Submodel]/[Add Submodel] took years to be added to
the spec, and usually by the time they appeared in a new version of the spec,
the
industry was already on to something new. IBIS just could never keep up,
because
it was inflexible, it needed new keywords and definitions to support the new
behaviors of the new buffer types.
I was concerned already when we were discussing the new IBIS-AMI additions to
the spec because it followed the same approach of predefined footprints, flows,
etc… If a certain (round) model or simulation does fit the (square) hole, you
are
out of luck. We are now starting to “reap the benefits” of this approach…
While
IBIS-AMI gave new life to IBIS because it did provide a solution to a need at
the
time, our inflexible implementation of that solution is starting run out of
steam
again, and we don’t know how to address it because no matter how we look at it,
it would take too much work and change in the spec…
Looking at these discussions, and the some of the new challenges we haven’t even
started to discuss yet, I am seriously concerned about the future or usefulness
of
IBIS-AMI… How are we going to go on from here?
Thanks,
Arpad
================================================================
From: Bob Miller (APD) [mailto:bob.miller@xxxxxxxxxxxx]
Sent: Monday, June 19, 2017 3:05 PM
To: Ambrish Varma <ambrishv@xxxxxxxxxxx>
Cc: wkatz@xxxxxxxxxx; Muranyi, Arpad <Arpad_Muranyi@xxxxxxxxxx>;
ibis-macro@xxxxxxxxxxxxx
Subject: Re: [ibis-macro] Re: Response to BIRD-190
If this is the case then I think I can safely say that IBIS-AMI does a poor job
of support for statistical simulation, not to even supply a truly workable path
for adaptation therein, or even a statistical-after-time-domain flow to patch
around it.
Regards,
Bob
