Re: [SI-LIST] : receiver jitter

About this list Date view Thread view Subject view Author view

From: Adrian Shiner (adrian.shiner@virgin.net)
Date: Thu Jan 20 2000 - 11:37:11 PST


Speaking as a user of these overgrown abacusses.

Let us remember that size & market share does not mean that a company REALLY
knows what is best.
There are many examples in history of companies hitting the rocks in pursuit
of the folly that they think they are untouchable. British Leyland in the UK
was one (car manufacturer). Now I saw a report in the Finanacial Times (that
pink paper) about a company who has been doing some "skunk works" on
developing something to give some bigger manufacturers of computing
semiconductors reasons to think again. If this situation develops as I hope
it does then a certain virtual duopoly will be broken at last. Linux will
help the process.

What we need now is more technical tools (rather than office tools) which
will run on Linux. Come on you simulator guys, help break the mould!

Adrian

----- Original Message -----
From: Jonathan Dowling <jdowlin2@yahoo.com>
To: <si-list@silab.eng.sun.com>
Sent: 19 January 2000 18:26
Subject: RE: [SI-LIST] : receiver jitter

>
> I'll tell you what.....Let's make a deal !!
> In the spirit of Andy Rooney's offer of $1,000,000.00 to anyone who can
identify Nicole's
> killer....
>
> In America, we vote with dollars. Instead of arguing, lets put it to such
a vote.
>
> Given that Intel is the bellweather of the industry and given that the
membership of the
> SI-LIST is dominated by Intel employees (as Ray can attest), the tool
adoption trend
> seen there should more or less be reflected in the majority of the
industry.
>
> I'm willing to concede the point when Intel can definitively report that
it has less than
> 100 IBIS compatible simulator licenses. This means all XTK, ICX,
SpecctraQuest
> licenses and other IBIS-like simulators. This includes those of you who
use linear excitation for
> MDS/ADS and Speed97. Consider yourselves behaviorists. Those who use the
behavioral
> modeling facility in Hspice should also be included.
>
> My best guess is that they have:
> 60-75 XTK
> 40-60 SpecctraQuest
> 5-10 ICX
> 25-50 Other
>
> If Intel does have more than 100 IBIS compatible simulator licenses, how
does this square with
> "the death of IBIS" scenario?
>
> Argueably, the worst problem with behavioral simulators is that they do
not (or poorly) predict
> switching noise. The trend to convert to differential signaling
eliminates this problem by
> suppressing SSO by 20x. This should enable a renaissance in behavioral
modeling and
> should increase its adoption rather than serve as a restriction.
>
> Interconnect researchers using Spice may want to find out who is
bankrolling their research
> these days. It may not say so on your paycheck, but while you fight
convergence problems, someone
> using a behavioral simulator is paying your bills through the design of
real, timely products
> that make real money.
> Right Chris?
>
>
> Jonathan Dowling
>
>
>
>
> --- Chris Cheng <hycheng@3pardata.com> wrote:
> > hmmm.... interesting. i wonder who design gtl buses first ?
> > ahh.... its got to be mr. gunning cause its called gunning tranceiver
> > logic.
> > ohhh.... he designed it first with a company call what ?
> > so.... please check who published the first system/bus design paper
> > in the industry on gtl. i give u a hint, the first author's name starts
> > with a c. and i assure u he only used spice to design the first gtl
> > system with 10 slot loading at 66MHz without a dead cycle in the
> > days of 80486.
> > for the record, i defined the corp. cad signal intrigity methodology
> > in sun more than 10 years ago and for sure behavoral model was in
> > it but over time performance requirement and accuarcy concern has
> > shift everything to spice. yes, i am still using behavoral model
> > when i am designing pci or jtag but it stops there.
> > actually intel has boat loads of spice license too (its free since
> > its in house tool). so i don't understand your comment on 100's of
> > behavoral simulators.
> > chris
> >
> > -----Original Message-----
> > From: owner-si-list@silab.eng.sun.com
> > [mailto:owner-si-list@silab.eng.sun.com]On Behalf Of Jonathan Dowling
> > Sent: Tuesday, January 18, 2000 5:57 PM
> > To: si-list@silab.eng.sun.com
> > Subject: Re: [SI-LIST] : receiver jitter
> >
> >
> > Newbies: Time for a history lesson...
> >
> > I would be willing to bet Chris' sentiments today were issued
> > verbatim by critics of behavioral modeling at the folks who
> > developed behavioral simulators in the 1980's (or before).
> > Proponents of SPICE have long criticized IBIS-like simulators,
> > even when signal quality not an issue (only timing).
> >
> > If we go back in time we would hear:
> > "We are going to be running at 4MHz this year. You can't correctly
> > predict timing with a behavioral model. This is 5V logic. You will
> > have to use SPICE. Specifically, you can't model the receiver well
> > enough."
> >
> > The true test is time-to-market. Almost without exception, all
> > the companies that excel at time-to-market in the digital world
> > have boatloads of behavioral simulator licenses. Intel is the
> > best example of this fact because they make the most money.
> > Their behavioral simulator licenses number in the 100's. If you
> > claim your company excels at time-to-market without behavioral
> > simulation, I claim your company is not as successful as Intel.
> >
> > Rambus's SPICE bigotry certainly contributed to their poor
> > time-to-market showing last year. Maybe they don't have a single
> > behavioral simulator in-house. They were generous enough to make
> > an IBIS model, but it was pooh-poohed by the 'experts' who claimed
> > (on this reflector list) that only the SPICE model would do.
> > Simulation with the simple IBIS model is enough to show timing
> > marginality at 800MT/s. (By the way, it looks great at 600MT/s
> > if you're lucky enough to avoid the resonances.)
> >
> > Jonathan Dowling
> >
> >
> >
> >
> >
> > --- Chris Cheng <hycheng@3pardata.com> wrote:
> > > as core and signal swing getting smaller and smaller, this
> > > problem is getting worst. compound this with source sync
> > > buses where the strobes and data can behave differently
> > > (true differential vs. pseudo differential), i believe
> > > predicting the response of the receiver under different
> > > waveforms input will be as important as simulating
> > > the propagation of signal in the interconnect
> > > environment. this will be the final nail in the coffin for
> > > behavioral models like ibis which is totally incapable of
> > > handling such analysis. i have seen pitiful attempts to
> > > qualitatively describe it in ring back or edge rate
> > > extrapolation but without the ability to simulate the
> > > receiver behavior, those specs are just fancy words with
> > > no meaningful insight into the response of the true
> > > receiver.
> > > can i spell SPICE again.
> > > chris
> > >
> >
> > __________________________________________________
> > Do You Yahoo!?
> > Talk to your friends online with Yahoo! Messenger.
> > http://im.yahoo.com
> >
> > **** To unsubscribe from si-list: send e-mail to
> > majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> > si-list, for more help, put HELP.
> > si-list archives are accessible at http://www.qsl.net/wb6tpu
> > ****
> >
> >
> >
> > **** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of
> > message put: UNSUBSCRIBE si-list, for more help, put HELP.
> > si-list archives are accessible at http://www.qsl.net/wb6tpu
> > ****
> >
> >
> __________________________________________________
> Do You Yahoo!?
> Talk to your friends online with Yahoo! Messenger.
> http://im.yahoo.com
>
> **** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu
> ****
>

**** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Apr 20 2000 - 11:34:45 PDT