FreeLists / si-list / [SI-LIST] Re: PCI Energy ...

[Scott McMorrow <scott@xxxxxxxxxxxxxxxx>]

All,

The PCI bus was most certainly not designed to be "green".  The prime
objective was to be easy (cheap) to fabricate with early 90's semiconductor
processing.   This ment reflective wave switching for minimum output transistor
size and diode clamping and no source series termination.  Adding source
series termination would have required a larger output drive strength and
thus larger transistor sizing.

Since the bus is reflective wave, the energy required will be determined
by the settling time of the bus, which depends upon the length of the
bus and it's loading.  Since the bus does not allow for source series 
termination, it is
possible that a PCI bus will never settle from one cycle to the next.(Limited, 
of course,
by data patterns.)   There are many busses in operation which have this 
behavior.

A bus with series termination would have lower worst case and average
power dissipation than the PCI bus has.  And, with carefully chosen
and placed series attenuators in the bus,  the overall power dissipation
can be decreased at the same time as signal integrity and timing
quality is improved.  We have done this many times with "huge"
PCI busses.

If the bus were are "perfect" reflective wave switching bus with source
termination, there would be a 50% improvement in power dissipation over
a parallel terminated bus for the AC switching condition, and a 100%
improvement in the DC condition.  But, since the bus is not a perfect
reflective wave switching bus, and does not settle quickly, the actual
worst case AC improvement in power dissipation is much less than
50%, and in the worst case may actually be exactly the same as the
parallel terminated bus (i.e. no settling between cycles.)  So what is
then left is the DC power savings.

regards,

scott


--
Scott McMorrow
Principal Engineer
SiQual, Signal Quality Engineering
18735 SW Boones Ferry Road
Tualatin, OR  97062-3090
(503) 885-1231
http://www.siqual.com



Doug McKean wrote:

> "Michael Nudelman" wrote:
> >
> > Now, if the "green" architecture is behind all this, tell me, how
> much
> > energy on the whole computer scale (percent-wise) do you really save
> on the
> > PCI data transfer? And is it worth it to complicate the design?
>
> That's an interesting question because all this time I 'thought' I
> knew
> the answer.  Since it's reflective based signaling, it's additive
> nature
> makes the final wave double bringing up to an acceptable level.
> So, the energy saving, I thought. was near 50%.
>
> Think I've got that all wrong now.
>
> Comments?
>
> - Doug
>
> ------------------------------------------------------------------
> To unsubscribe from si-list:
> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
> List archives are viewable at:
>                 http://www.freelists.org/archives/si-list
> or at our remote archives:
>                 http://groups.yahoo.com/group/si-list/messages
> Old (prior to June 6, 2001) list archives are viewable at:
>                 http://www.qsl.net/wb6tpu
>





------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages 
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu