[ibis-macro] Meeting minutes - 7/18/2006
- From: "Todd Westerhoff \(twesterh\)" <twesterh@xxxxxxxxx>
- To: <ibis-macro@xxxxxxxxxxxxx>
- Date: Tue, 18 Jul 2006 16:55:22 -0400
Meeting date: 18 jul 2006
Members (asterisk for those attending):
*Arpad Muranyi, Intel Corp.
*Bob Ross, Teraspeed Consulting Group
*Todd Westerhoff, Cisco Systems
Mike LaBonte, Cisco Systems
Paul Fernando, NCSU
*Barry Katz, SiSoft
*Walter Katz, SiSoft
Ken Willis, Cadence Design Systems
*Ian Dodd, Mentor Graphics
*Lance Wang, Cadence Design Systems
Richard Ward, Texas Instruments
Doug White, Cisco Systems
Sanjeev Gupta, Agilent
Joe Abler, IBM
John Shields, Mentor Graphics
Review of ARs:
- Arpad, Mike & Todd to discuss presentation for IBIS Summit
==> Done
- Arpad to contact Paul Fernando
==> Done, Paul can continue to work on project
-------------
Todd evidently messed up the meeting invitation; Cisco MeetingPlace
wouldn't let anyone into the Web portion of the meeting
(except Todd, but a web meeting for one isn't all that useful).
Discussion of circuit simulation versus signal processing
- circuit simulators model voltage and current iteratively as the time
wheel advances
- the API proposed by Cadence seems have the input stimulus defined for
all time when the model is invoked
(i.e. Kirchoff's laws are not solved, currents are not predicted)
Operative question is - does the Cadence API proposal fit in with the
current level of simulation as defined by IBIS?
- Discussion of what constitutes a linear or non-linear circuit
- We noted that in a SERDES receiver, the input buffer effectively
separates the internal logic from the channel
behavior. Therefore, no matter what algorithm is applied to clean up
and recover the input, the channel
behavior remains the same.
- In the case of receiver algorithm modeling, it becomes apparent that
as long as the input to the receiver is
known, the algorithm can be applied to that pattern. It doesn't
matter, for example, whether the receiver
input was derived from fingerprinting a channel and performing
convolution, or conventional time domain
simulation - once the signal at the receiver is known, the algorithm
can be applied to it.
- What happens when the circuit behavior changes - for instance, if the
receiver adjusts its termination setting
to improve channel characteristics, or if the driver changes its tap
settings? How is the change in the
channel's characteristics handled?
- What happens if we model a driver with a complex compensation
algorithm, for instance, one that changes its tap
settings based on the local data pattern? In the Cadence approach,
do we need a channel fingerprint for each driver
TAP setting?
-------------
AR: (Cadence) Need more technical detail from Cadence, Kumar to
attend next/future meetings
AR: (Intel/Cisco) Arpad, Todd, Mike to complete presentation for DAC
IBIS Summit
No meeting next week due to DAC IBIS Summit
- Arpad on vacation 1st week of Aug
- Todd on vacation 1st two weeks of August
Next meeting: Tuesday 1 Aug 2006 12:00pm PT
Other related posts:
