"Pentium 4" brandname ready to be dropped

[Yousuf Khan]

Reet petite and awesome neat

What I'd love to see those in is all my battery-powered
tools - especially for outdoor use in the winter.

And I'd love to see all of those clouds of diesel fumes trailing people as
they use their cellphones.

Yousuf Khan

 

[Rob Stow]

And I'd love to see all of those clouds of diesel fumes trailing people as
they use their cellphones.

Yousuf Khan

Sales of Chanel #5 are about to take a huge boost ?

 

[Mike Smith]

In the section you snipped you missed the part about
wanting an ATX motherboard.

The AOpen one is just another one of those
less-than-full-featured micros. Only 2 DIMM
slots and only 3 PCI slots just doesn't cut it -
particularly when it costs twice as much as a
full-featured ATX board.

Indeed. If I were really desperate, I could buy one of those mini-ITX
jobs from Commell or Lippert, but damn are they expensive.

A lot of people - but
not me - would also be disappointed by no AGP 3.0.

I could go either way. I would love to build a Pentium M gaming system,
but what I'm really looking for is to be able to build low-power,
low-noise (preferably silent) servers to experiment with. (I live in a
small apartment and don't have the luxury of a spare room where I can
put all the computers and then keep the door closed.)

 

[Nick Maclaren]

|> > Rob Stow wrote:
|> >
|> >>>You may not need a nuclear reactor to power these things, but soon
|> >>>you may be able to get microscopic diesel-fueled jet engines to
|> >>>power them. http://www.newscientist.com/news/news.jsp?id=ns99996559
|> >>
|> >>Reet petite and awesome neat
|> >>
|> >>What I'd love to see those in is all my battery-powered
|> >>tools - especially for outdoor use in the winter.
|> >
|> > And I'd love to see all of those clouds of diesel fumes trailing people as
|> > they use their cellphones.
|>
|> Sales of Chanel #5 are about to take a huge boost ?

You have forgotten the need to water-cool them - it's really
Channel number five.


Regards,
Nick Maclaren.

 

[Stephen Fuld]

snip

What I'd love to see those in is all my battery-powered
tools - especially for outdoor use in the winter.

For those, you might not want to have the jet engine run a generator to
create electricity. It might make more sense to gear down the turbine to
turn the shaft on the tool directly (i.e. like a prop jet) But I think
these are a long way off.

 

[Johannes H Andersen]

http://www.theinquirer.net/?article=19187

Yousuf Khan

Will the next one be an Intel Pentium Pentium, i.e. a Pentium 5?

 

[keith]

And I'd love to see all of those clouds of diesel fumes trailing people as
they use their cellphones.

Hmm. It would make Sidewinders more effective (in both Afganistan and
here.

 

[Stephen Sprunk]

It's likely that all of the dual cores whether from AMD or Intel will be a
couple frequency steps behind their single-core cousins. I think the only
question is how many frequency steps behind they will be, and whether one
mfg or the other will be able to keep the percentage of the drop lower
than the other (eg. one might only be one step behind their own
single-core, while the other one might be two steps behind).

http://www.theinquirer.net/?article=18967

"At MPF this week, AMD said two things of note, clock speed and performance.
The clock speed was given as three or five speed grades below the prevailing
chip at the time. This fits in well with the reports I am getting that it
will basically be a couple of low voltage cores on a slice of silicon to
remain under the 95W power cap."

"The performance estimates AMD gave fit even more closely. They were saying
that the dual core chips will be between 130 and 160% of the performance of
the highest clocked single core."


I haven't seen any similar info on the relative speed of Intel's dual-core
chips, but that's not surprising with the recent flurry of roadmap updates.

S

 

[Tony Hill]

Such as ? Take the AMD64 processors, for example.
Multi-core would be the first significant change to
the AMD64 architecture since the Opty 140 and 240 were
released at 1.4 GHz. All we have seen in the meantime
is a steady jacking up of clock speeds and there is
nothing else on the horizon for the next 6 to 9 months.

Come now, they've only been out for a year and a half! AMD has made a
few tweaks to the Athlon64/Opteron core with the 90nm shrink, though
nothing major.

On the Athlon64 side of things AMD has changed the socket used to
increase memory bandwidth (or reduce cost relative to Socket 940) and
added Cool 'n Quiet to reduce power consumption.

The situation has been much the same for the P4 since
it first came out. Many small changes have been made
to allow Intel to keep jacking up clock speeds, but
the basic chip design has stayed the same.

Err, except that the cache has gone from 256K up to 1MB (2MB for the
Extremely Expensive Edition and maybe future standard-P4s), the bus
speed has gone from 400MT/s up to 800MT/s (with 1066MT/s being a
possibility in the near future), the core was significantly redesigned
for the Prescott (albeit with rather unimpressive results). SSE3 was
added in and 64-bit support is available in limited situations (only
to OEMs at the moment).

I would say that they have made some changes. Maybe not all the right
changes, but they have made changes.

You could make an argument for the Pentium M as being
Intel's effort to get performance at lower clocks and
without needing a nuclear reactor in every home, but
since you *still* can't buy a full-fledged ATX
motherboard for Pentium M the point is pretty much moot.

Yes, rather disappointing, those Pentium-Ms are rather nice little
chips but Intel really just doesn't want people using them in desktop
systems. I have no idea why not, it doesn't seem to make any sense to
me, but I'm sure they've got some misguided reasoning.

 

[Rob Stow]

Come now, they've only been out for a year and a half! AMD has made a
few tweaks to the Athlon64/Opteron core with the 90nm shrink, though
nothing major.

True enough - and in that year and a half all AMD
has really done with AMD64 processors is to jack up
the clock speeds. I'm *not* putting them down for that.
My whole point is simply that they can't keep doing
that indefinitely - which is why they are switching to
dual cores as another way to keep jacking up performance.

The first paragraph I posted in this thread hasn't
been snipped yet, but what came immediately before
it /has/ been snipped. The poster I was replying to
was lamenting the fact that dual cores are apparently
going to be accompanied by lower clocks. And I say
again: isn't that the whole point of dual cores ?
Clocks simply can't keep going up the way they have in
the past.

On the Athlon64 side of things AMD has changed the socket used to

AMD has not really changed the socket. All
three socket types were planned for long before
the first Opterons were released. For some reason
they put off releasing the socket 939 versions for
a long time, but it was always something they had
intended to do.

increase memory bandwidth (or reduce cost relative to Socket 940) and
added Cool 'n Quiet to reduce power consumption.

I was under the impression that C & Q is was a feature
that has been in all of the chips since day one, but
disabled because it had bugs in it that needed to be
worked out.

 

[=?ISO-8859-1?Q?Jan_Vorbr=FCggen?=]

For those, you might not want to have the jet engine run a generator to

create electricity. It might make more sense to gear down the turbine to
turn the shaft on the tool directly (i.e. like a prop jet)

No, for tool use you want to have an electric motor driving the shaft -
it has much better properties than any other engine for this purpose (think
stalls and low-speed behaviour, for instance). Most (all?) Diesel-powered
train engines are really Diesel-electric.

Jan

 

[Douglas Siebert]

True enough - and in that year and a half all AMD
has really done with AMD64 processors is to jack up
the clock speeds. I'm *not* putting them down for that.
My whole point is simply that they can't keep doing
that indefinitely - which is why they are switching to
dual cores as another way to keep jacking up performance.

The first paragraph I posted in this thread hasn't
been snipped yet, but what came immediately before
it /has/ been snipped. The poster I was replying to
was lamenting the fact that dual cores are apparently
going to be accompanied by lower clocks. And I say
again: isn't that the whole point of dual cores ?
Clocks simply can't keep going up the way they have in
the past.

Why do you assume clocks can't keep going up the way they did in the
past? Just because Intel reached a limit with Prescott, which may be
due to a lot of reasons other than "scaling is dead", you can't assume
the same is true for AMD, or Intel's other designs. I'm not making
any claims one way or another, just that you are drawing conclusions
on limited data. Certainly there's no evidence AMD believes it can't
scale MHz effectively in their 90nm process, and no reason to believe
that either Intel or AMD couldn't do so in their 65nm generation as
well.

IMHO the bigger driver for dual cores is that smaller processes
allow for more transistors, and the easiest ways now to use those extra
transistors are for bigger cache or for additional cores. In the past
you'd see processors made more complex, by widing the superscalarity.
3 way x86 is at now there is a point of diminishing returns. Though I
think Intel is rumored to have a 4 way core in their next x86 design...

There are certain things that drive a "sweet spot" for die sizes. You
can only make them so small before you run into packaging issues, and
smaller dies result in more chips which if you can't sell them all
leads to unsold inventory that you have to write down -- something Intel
is currently going through. That's probably one of the reasons why
Intel is increasing the cache size to 2MB, and moving to dual cores on
the desktop (since it is higher volume and can soak up more capacity)
before servers.

This is in contrast to AMD, which is doing dual cores on the server
first, where the larger profit is, since they do not have oversupply
issues, but a much larger desktop die size could potentially lead them
into an undersupply. Look for dual core desktop Athlon 64s to get
pushed forward if AMD starts having a lot of spare capacity (either due
to lower sales in the overall CPU market, or AMD's market share staying
stagnant or declining) If AMD's market share starts to shoot through
the roof, look for the dual core desktop Athlon 64s to get pushed back
to the 65nm generation.

 

[Yousuf Khan]

Why do you assume clocks can't keep going up the way they did in the
past? Just because Intel reached a limit with Prescott, which may be
due to a lot of reasons other than "scaling is dead", you can't assume
the same is true for AMD, or Intel's other designs. I'm not making
any claims one way or another, just that you are drawing conclusions
on limited data. Certainly there's no evidence AMD believes it can't
scale MHz effectively in their 90nm process, and no reason to believe
that either Intel or AMD couldn't do so in their 65nm generation as
well.

Personally, I think we'll see 4Ghz again in maybe another two or three
years. It just won't be from a Pentium-4-derived processor, which I'm
assuming will be history by then.

Yousuf Khan

 

[Bill Davidsen]

Indeed. If I were really desperate, I could buy one of those mini-ITX
jobs from Commell or Lippert, but damn are they expensive.



I could go either way. I would love to build a Pentium M gaming system,
but what I'm really looking for is to be able to build low-power,
low-noise (preferably silent) servers to experiment with. (I live in a
small apartment and don't have the luxury of a spare room where I can
put all the computers and then keep the door closed.)

I built a SHuttle system, something-51 case, and I put in a 5400 rpm
drive and 2GHz Celeron. The fan almost never runs, the drive is quiet,
and I think I have about $300 invested in parts I bought (new) at a
local computer show. The heat pipes on the Shuttle cool the CPU with the
fan off much of the time.

Note that quiet is usually low power, and low power is not normally a
gaming system. PM is ideal, but pricy.

Did you consider a year-old used laptop? Gets most of what you want and
a UPS built-in.

 

[Rob Stow]

Why do you assume clocks can't keep going up the way they did in the
past?

Because we are soon to be hitting physical limits
in how small things can be and how much power we
can pump through those itty bitty things without
melting them.

Up until now and for the near future, the physical
limits haven't been what was holding us back - it
was our manufacturing technology. Now manufacturing
technology is just about caught up to the physical
limits, so what will be left except to manufacture
bigger or better chips instead of chips that are
merely clocked faster ?

You might want to take a look at what Keith Williams
had to say on Monday in the thread
"Re: Processor heat dissipation, Leakage current, voltages & clockspeed"

Here's part of it ...
<quote typos=fixed>
Face the facts. *We* are getting perilously close
to atomic dimensions and the voltage gradients are
constantly flirting with the MV/cm "limit". *We*
now have 100A on a chip, not much bigger across than
the wire supplying power to your electric stove ...
and the current is all on the "surface". The power
density of these things are on the order of a *BILLION*
times that of ol' Sol.

Just because Intel reached a limit with Prescott, which may be
due to a lot of reasons other than "scaling is dead", you can't assume
the same is true for AMD, or Intel's other designs. I'm not making
any claims one way or another, just that you are drawing conclusions
on limited data. Certainly there's no evidence AMD believes it can't
scale MHz effectively in their 90nm process, and no reason to believe
that either Intel or AMD couldn't do so in their 65nm generation as
well.

AMD is currently at 2.4 GHz with their 90 nm chips, and
the two that I've had my hands on have had little room
for overclocking - even when the fan is blowing -10'C
air over the heatsink.

Do I think AMD is going to solve this and get up to the
kinds of clocks Intel hit with the P4 ? Yes - eventually.
Do I think 65 nm is going to be a magic cure-all that's
going to burst through the 3.6 GHz ceiling Intel hit ?
I'll believe it when I see it.

 

[Anton Ertl]

Why do you assume clocks can't keep going up the way they did in the
past?

Because they have not kept going up last time the way the did earlier.
Up to the 0.18u generation Intel and AMD have very reliably produced
clock speedups by factors of around 2 per generation of the same core.
With the 0.18u-0.13u they only produced speedups by a factor 1.7 or
less (Williamette 2GHz -> Northwood 3.4GHz, Thunderbird 1.4GHz ->
Thoroughbred 2.25GHz, and that's not counting the Palomino (~1.7GHz)
as 0.18u chip).

That's probably one of the reasons why
Intel is increasing the cache size to 2MB

Also, Intel seems to have a trick for fitting lots of cache in little
space die that AMD does not have: 0.5MB of cache take 49mm^2 on 0.13u
AMD CPUs (Clawhammer-Newcastle differences); I would expect that in
90nm 1MB of cache take about the same area. However, Dothan's cache
takes about 42mm^2 (about half of the 84mm^2 die) and has 2MB;
Prescott 1MB cache also looks small.

Followups to comp.arch.

- anton

 

[Mike Smith]

I built a SHuttle system, something-51 case, and I put in a 5400 rpm
drive and 2GHz Celeron. The fan almost never runs, the drive is quiet,
and I think I have about $300 invested in parts I bought (new) at a
local computer show. The heat pipes on the Shuttle cool the CPU with the
fan off much of the time.

Tried the XPC route, though with an Athlon XP 2000+. It would probably
be OK if sufficiently well hidden, but as a personal machine I found it
too noisy. Now, this was an SK41G, one of the early ones, so it's quite
possible that they've gotten better since then.

Note that quiet is usually low power, and low power is not normally a
gaming system.

Thus my dilemma. ;-)

PM is ideal, but pricy.

Pricey - *within reason* is OK.

Did you consider a year-old used laptop? Gets most of what you want and
a UPS built-in.

For the machine I sit in front of, that's exactly the route I chose -
right now I have a P4M 1.8 GHz w/ GeForce 440 Go (coupla years old), and
I am about to migrate to a Dothan 1.7 w/ Radeon 9600. However, even on
that machine I'm already discovering that Doom 3 has its slow moments.
What I need is a Dothan notebook with MR9800 or GeForce 6600 Go or
something, but AFAIK there ain't no such animal.

For the server boxes (nothing heavy-duty, just something I can play with
to learn Linux and Apache and MySQL and such) for the time being I'm
going the mini-ITX route, with VIA boards and cases from Logic Supply
(http://www.logicsupply.com) - they specialize in mini-ITX and
small/quiet stuff.

I may just end up building a separate game machine, and forget about
trying to get it all in one package. Then maybe I'll just duct-tape a
vacuum cleaner hose to the case for cooling. ;-)

Thanks for comments,

 

[Rob Stow]

Tried the XPC route, though with an Athlon XP 2000+. It would probably
be OK if sufficiently well hidden, but as a personal machine I found it
too noisy. Now, this was an SK41G, one of the early ones, so it's quite
possible that they've gotten better since then.



Thus my dilemma. ;-)



Pricey - *within reason* is OK.



For the machine I sit in front of, that's exactly the route I chose -
right now I have a P4M 1.8 GHz w/ GeForce 440 Go (coupla years old), and
I am about to migrate to a Dothan 1.7 w/ Radeon 9600. However, even on
that machine I'm already discovering that Doom 3 has its slow moments.
What I need is a Dothan notebook with MR9800 or GeForce 6600 Go or
something, but AFAIK there ain't no such animal.

For the server boxes (nothing heavy-duty, just something I can play with
to learn Linux and Apache and MySQL and such) for the time being I'm
going the mini-ITX route, with VIA boards and cases from Logic Supply
(http://www.logicsupply.com) - they specialize in mini-ITX and
small/quiet stuff.

I may just end up building a separate game machine, and forget about
trying to get it all in one package. Then maybe I'll just duct-tape a
vacuum cleaner hose to the case for cooling. ;-)

Ducting can make a *huge* difference.

I recently modded a system to use 2.5" flexible rubber
hose (from a swimming pool vacuum) and two 1800 rpm 60 mm
fans to provide ducted cooling for an Athlon64 3200+.
Having the ducting draw cool air in from outside the
case, blow it over the heat sink, then blow it outside
the case through another duct provides virtually silent
cooling for the CPU. Cooling the cpu with 20'C air
drawn from outside the case instead of with warm air
drawn from inside the case means you require a lot less
air to cool the cpu.

This is the second time I have done that kind of ducting
for cpu cooling and it worked great both times.

In addition, by having the heat from the CPU ducted out
of the case, the need for the two case fans was reduced
so much that I could slow one down to an inaudible
1200 rpm and turn off the other one altogether. If it
wasn't for the Radeon 9800 Pro in that system I wouldn't
have hesitated to simply turn off both case fans completely.

After I turned that system back over to the owner ( usual
friend of a friend situation ) he liked the noise reduction
so much that we did something similar to cool the PSU.
Now that the PSU is cooled by room temperature air instead
of with warm air from inside the case, the PSU fan speed
was cut in half - with the obvious resultant noise reductions.
This necessiated the turning the second case fan back on, but
since it too can be run at only 1200 rpm there is still a
noticeable improvement in noise levels from the PSU mod.

The downside is that the pre-mod fans were much
faster and noiser and they did a lot to mask the noise
made by other things - particularly the hard disks and
the DVD drive. Those things and things like keyboard
noise now seem much louder now the fan noise has been
reduced so much.

 

[Douglas Siebert]

Because we are soon to be hitting physical limits
in how small things can be and how much power we
can pump through those itty bitty things without
melting them.

Up until now and for the near future, the physical
limits haven't been what was holding us back - it
was our manufacturing technology. Now manufacturing
technology is just about caught up to the physical
limits, so what will be left except to manufacture
bigger or better chips instead of chips that are
merely clocked faster ?

Ah, sounds like "Death of the Internet predicted, film at 11pm"...

We're an over of magnitude away from hitting the limit in terms of "how
small things can be". Though its possible quantum effects might hurt
us before then, but we might find a way to harness them for further
improvements. I remember 10 years ago when it was fashionable to claim
that we were nearing the physical limits using optics and we'd need to
be using xrays or electron beams by now. But damned if those smart
boffins didn't find a way around that with new materials with amazing
refractive indices so we can continue using optics to the 45nm and
perhaps even the 32nm generation. Its even possible that by then we
might have new materials that ever better to extend the life of optics
even further.

As for power density limits, we are still in the caveman stages when
it comes to handling stuff like that, blowing air over pieces of metal
pressed down on the chip surface...c'mon! There are plenty of ways to
help with the cooling of localized heat buildup, from simple liquid
cooling like overclockers (and Apple) are doing now, to integrated
Peltier's (AMD's patent from 2001 that was recently publicized) to
stuff that's further out, like routing thousands of microscopic channels
through the chip for cooling fluid flow.

True, wires won't shrink as fast as feature sizes (and I think that's
been true for a while now and hasn't hurt us) That just means more
metal layers as we get smaller and more complex routing -- luckily we
have faster computers to help with that more complex routing.

AMD is currently at 2.4 GHz with their 90 nm chips, and
the two that I've had my hands on have had little room
for overclocking - even when the fan is blowing -10'C
air over the heatsink.

Yes, and this is their very first batch of 90nm chips. I remember a
lot of the same complaints about the first 130nm K8s that were shipped
being unable to exceed 2 GHz and there was a lot of worrying about
AMD's SOI process, but since they are shipping some parts at 2.6GHz in
130nm now it seems like they licked that problem pretty well.

 

[David Kanter]

Why do you assume clocks can't keep going up the way they did in the
past? Just because Intel reached a limit with Prescott, which may be
due to a lot of reasons other than "scaling is dead", you can't assume
the same is true for AMD, or Intel's other designs. I'm not making
any claims one way or another, just that you are drawing conclusions
on limited data. Certainly there's no evidence AMD believes it can't
scale MHz effectively in their 90nm process, and no reason to believe
that either Intel or AMD couldn't do so in their 65nm generation as
well.

Well, their going dual core, that should be a clue.

IMHO the bigger driver for dual cores is that smaller processes
allow for more transistors, and the easiest ways now to use those extra
transistors are for bigger cache or for additional cores.

Let me give you a hint, 200mm^2 cores won't WORK in 90nm...

You can spend transistors in many fashions:
1. Improving the core
2. Improving the cache
3. Replicating the core

1 causes problems if you use too much area. 2 is pretty inexpensive,
from the area, defect and design standpoint, and 3 is inexpensive from
the design standpoint (I suspect).

Large area cores take too long for signals to traverse and are also
problems from the yield stand point.

In the past
you'd see processors made more complex, by widing the superscalarity.
3 way x86 is at now there is a point of diminishing returns. Though I
think Intel is rumored to have a 4 way core in their next x86 design...
There are certain things that drive a "sweet spot" for die sizes. You
can only make them so small before you run into packaging issues, and
smaller dies result in more chips which if you can't sell them all
leads to unsold inventory that you have to write down -- something Intel
is currently going through.

Well, you also cannot make cores too large or it takes too long for
signals to traverse the core. Thats the major restriction to
uber-wide OoO cores IMHO.

That's probably one of the reasons why
Intel is increasing the cache size to 2MB, and moving to dual cores on
the desktop (since it is higher volume and can soak up more capacity)
before servers.

Actually, IMHO that has more to do with heat and scaling issues
driving the need for more cache and the inventory issues make larger
die sizes more palatable.

BTW, I believe that Intel will be releasing Dual core Xeons at the
same time as dual core desktops.

This is in contrast to AMD, which is doing dual cores on the server
first, where the larger profit is, since they do not have oversupply
issues, but a much larger desktop die size could potentially lead them
into an undersupply.

That's a good point, as dual core on the desktop would halve AMD's
capacity. Maybe Intel should have thought about going quadcore...that
would sure put some pressure on AMD.

Did AMD state that they would release a dual core server chip before
the desktop, or do they simply not intend to release a dual core
desktop chip?

Sorry, I stopped caring about the desktop market a while ago...

Look for dual core desktop Athlon 64s to get
pushed forward if AMD starts having a lot of spare capacity (either due
to lower sales in the overall CPU market, or AMD's market share staying
stagnant or declining) If AMD's market share starts to shoot through
the roof, look for the dual core desktop Athlon 64s to get pushed back
to the 65nm generation.

I doubt this will happen. Intel can easily push dual cores as a huge
desktop advantage (twice as good). It will show up in benchmarks,
unlike HT, which was just a blip. I would expect to hit ~30-70% gains
on certain benchmarks, and also the systems will be more responsive
since normal users have quite a few processes running at once.

David