[sv-dc] resolution function with a state ?

Hi,

following I describe a simple and common scenario for real-value or
logic resolution. I am not sure if our current resolution methodology
can handle this efficiently.

Lets consider logic values first:
we have a driving gate A (VDD=1.2V), two loading gates B (VDD=1.8V) and
C (VDD=1.0V) and a common net "n".
Lets assume "A" has driven a "0" for a while and suddenly drives a "1".
"A" shall have a driving strength of 1/10k(Ohm) and "B" and "C" load
with altogether 100f(F).

In that case the RC-constant of the common net is 10kOhm*100fF = 1ns and
thus the slew rate is 1.2V/1ns = 1.2GV/s.
A resolution module might simply schedule a "1" for "B" within
1.8/2/1.2G s = 0.75ns and a "1" for "C" in 1.0/2/1.2G s = 0.42ns.

A resolution function that is exclusively based on Norton/Thevenin
equivalents and has no intrinsic state and timing for "n" would have to
render this inertial behavior by a sufficiently large number of time
steps within the connected Norton/Thevenin equivalents (provided there
is a way to determine the appropriate time step for "n").

Basically the same applies if we take real-value drivers and
readers/loads instead of logic.

So the resolution MODULE would be more accurate and 10-100x faster than
a resolution FUNCTION. A resolution module would also allow to easily
attach an initial value to a net and it can be the right place to take
care of an optimum timing for triggering and reevaluating the
Norton/Thevenin equivalents.

These are some of the reasons, why I think a resolution MODULE with
intrinsic state and timing might be a better methodology for practical
application.

Please let me know about your opinion.
Thanks,
       Achim

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.