Intel embarrassed by own report

[Robert Redelmeier]

I had two line a few years back. The original line wouldn't
go over 26K, while the newer one was reliably 53K.

Sounds like load coils on the old line.

Any discontinuity at a splice is peanuts at these
frequencies. That's like the people replacing their internal
house wiring with cat-5, expecting to see better perfromance.

And they usually won't. But push the distance and you also
get more splices. It doesn't help at 1.1 MHz.

No, I meant get rid of their land-lines completely.

Fine for short ranges or low bandwidth.

structures in the village. The loop isn't quite that old
though. ;-) He has no problem at 53K. Old wire isn't noiser
than new wire (assuming twisted pair, not the individual
wires on the tree).

My parents house is also much older, with older wiring they
only get 48k . That's still no load coils, but the line is
starting to limit. At 8kft, they probably could get DSL 1.5 .

-- Robert

 

[keith]

Sounds like load coils on the old line.

Nope, it went all the way back to the CO, rather than a "slick" in the
neighborhood.

And they usually won't. But push the distance and you also
get more splices. It doesn't help at 1.1 MHz.

Oh, we're talking about different things. I thougth we were talking POTS.
If you take the POTS DACS out of the line you can up the frequency, which
is the other data-rate knob to twist. POTS is stuck with the 8kHz DAC, so
the only available knob is S/N.

Fine for short ranges or low bandwidth.

Let me back up... As people ditch their phone lines, the rest become
quieter. ;-)/2

My parents house is also much older, with older wiring they only get 48k
. That's still no load coils, but the line is starting to limit. At
8kft, they probably could get DSL 1.5 .

Again, I thought we were talking about V.9x sorts of things.

 

[Robert Myers]

Why, certainly.

http://www.ibm.com/investor/news/jun-2005/03-06-05-2.phtml

Thanks. One piece of news there, of course, is that IBM is offering
its own tools (and, I suspect, your services) to the public.

That news says something about the importance of process variability.
It doesn't say how fast the problem is getting worse, and
electromigration is a problem on top of that.

Most of the scaling predictions we've seen (Moore's law will break
down at some particular scale) have turned out to be wrong, but sooner
or later, this (process variability, leakage, electromigration) all
has to catch up with us. I suppose a great many people would like to
know when.

RM

 

[Robert Redelmeier]

Nope, it went all the way back to the CO, rather than a
"slick" in the neighborhood.

Yes, but if it's a long or boraderline run, it could get
load coils.

Oh, we're talking about different things. I thougth we were
talking POTS. If you take the POTS DACS out of the line you can
up the frequency, which is the other data-rate knob to twist.

Yes. I was thinking specifically of DSL. For POTS, that
prof's estimate of 20k being max isn't really all that far
of 53k. Only a factor of 2.6

Let me back up... As people ditch their phone lines,
the rest become quieter. ;-)/2

Of course. Particularly if they aren't carrying modem
signals for hours on end.

-- Robert

 

[Hank Oredson]

Sounds like load coils on the old line.

Might be, might be a phantom or bridged line.
Telco will fix those things when you fuss at them.
Had to do that here ...

And they usually won't. But push the distance and you also
get more splices. It doesn't help at 1.1 MHz.


Fine for short ranges or low bandwidth.

Satellite works fine here, much better than the land line internet.

My parents house is also much older, with older wiring they
only get 48k . That's still no load coils, but the line is
starting to limit. At 8kft, they probably could get DSL 1.5 .

--

... Hank

http://home.earthlink.net/~horedson
http://home.earthlink.net/~w0rli

 

[Hank Oredson]

In the MHz for subscriber loop? Maybe for inter-office
trunking before it was mostly replaced by fiber.

The data is the characteristics of the loop.
Run it at whatever frequency you like.

--

... Hank

http://home.earthlink.net/~horedson
http://home.earthlink.net/~w0rli

 

[keith]

Yes, but if it's a long or boraderline run, it could get
load coils.

No load coials anywhere around, at least on the local loops. I asked.

Yes. I was thinking specifically of DSL. For POTS, that
prof's estimate of 20k being max isn't really all that far
of 53k. Only a factor of 2.6


Of course. Particularly if they aren't carrying modem
signals for hours on end.

;-)

 

[Del Cecchi]

Thanks. One piece of news there, of course, is that IBM is offering
its own tools (and, I suspect, your services) to the public.

That news says something about the importance of process variability.
It doesn't say how fast the problem is getting worse, and
electromigration is a problem on top of that.

Most of the scaling predictions we've seen (Moore's law will break
down at some particular scale) have turned out to be wrong, but sooner
or later, this (process variability, leakage, electromigration) all
has to catch up with us. I suppose a great many people would like to
know when.

RM

IBM has been offering the services of a goodly number of expert
designers at all levels from silicon wafer to system to software "to the
public" (assuming said public has enough money) for a few years now as
E&TS.

My personal opinion is that the problems you mention will slow progress
in a gradual fashion, rather than acting like a wall. The improvement
between generations will gradually become less, and the time span greater.

 

[Alexander Grigoriev]

The worst discontinuities are "split trunks" (?? I forgot exact words). I
mean, a trunk, connected from the middle of a live local loop. But DSL
modems can handle that.

 

[Keith R. Williams]

IBM has been offering the services of a goodly number of expert
designers at all levels from silicon wafer to system to software "to the
public" (assuming said public has enough money) for a few years now as
E&TS.

My personal opinion is that the problems you mention will slow progress
in a gradual fashion, rather than acting like a wall. The improvement
between generations will gradually become less, and the time span greater.

....and the cost higher. We've been seeing this for a couple of
generations now. ...increasing difficulty/cost with less payback.

 

[Keith R. Williams]

Might be, might be a phantom or bridged line.
Telco will fix those things when you fuss at them.
Had to do that here ...

No, that was the first thing I checked. It's just a long (~8km) noisy
line.

Satellite works fine here, much better than the land line internet.

Much expensive too. At least here they're about 5X the cost of cable,
plus about $1K in equipment. Better than nothing though (as he sits in
a hotel room where there are no local Internet POPs).