[lowem]

Reuters reports that :

Intel Corp on Thursday canceled plans to introduce its highest-speed desktop computer chip, ending for now a 25-year run that has seen the speeds of Intel's microprocessors increase by more than 750 times .... cranking up the speed of its Pentium 4 chips to four gigahertz, or billions of cycles per second, has been an elusive goal for Intel. When the latest rendition of the Pentium 4 was introduced in February, Intel said it would reach the 4 gigahertz speed by the end of the year.

... in July, citing concerns about having enough supply to meet customer demand, Intel delayed plans for four gigahertz until the end of March 2005. Thursday's announcement puts an end to the goal entirely, at least for the current generation of processors. The unexpected move also adds to a string of product changes, cancellations and recalls that has roiled the world's largest chip maker this year ...

- "Peak Chip", anyone? Is this *it* - *gasp* - the End of Moore's Law?

See also :

1. Anything to do with Moore?
2. The Collapse of Complex Societies (Joseph Tainter)
3. Energetic Limits to Growth (Jay Hanson)

(re-post from weblog)
http://www.jroller.com/page/lowem/20041 ... _4ghz_chip

[NevadaGhosts]

I thought you were talking about Doritos. Peak Doritos would be a serious problem.

[larrydallas]

This does not come as a surprise to me. I grew up in the 90's and as any child I was not aware of things following a bell curve. BUT; I would say I always had this feeling where I was bothered by how fast things were changing and how the newst stuff was out dated in a matter of months.

The 90's were just an era of progress and over development in all aspects. I'd say we packed in 50 years worth of consumption into that 10 years and now the price is being paid.

A very simple example would be the invention of the computer and applications to business. Say you own a car compnay and before you used to have about 500 accountants do the required accounting wiht adding machines and primtive computers of sorts.

You buy a program which allows you to lay off 490 of your people and now you make a one time investment in technology which saves you money in the long run; or does it?

The 490 people no longer have a living wage so they can not consume. They can not buy the cars you are producing. They can look for another job with other companies but odds are that compettion is so strong the other companies will also be doing the same thing.

Displaced workers can get into other industries but eventually all industries will run this way. So, this is when risky loans and dangerous investments is made by the individual and business so keep the money flowing. This is where we are today. This is a huge bubble waiting to pop.

The people at Intel cut their own throats. By replacing their own technology every 6-12 months they cut away at what would be possible to produce in the real world at too high a rate. BTW, how come pentium 3 processors were only around for about a year until P4 came out?

I can recall P2 being out years but P3 was short lived.

Anyway, it's notvery smart to think you are very smart when you totally dismiss the long term.

Instant gratification is a leading cause of the decline of our society.

[clv101]

They may be cancelling the P4 but only 'cos the architecture is at the end of the line. Next years chips will be more powerful, just they'll do it with an architecture that doesn't need 4GHz to get there. There's a lot of working going into Centrino style chips and asynchronous chips that don't have a hard and fast clock like today’s architecture.

[Barbara]

dallas,
the theory you mentioned is called "Fordism", because the great Henry Ford invented it. "Pay your workers a good wage so they can buy your products".
Now economists proudly announced we are in the post-fordism era.
Karl Marx said, more than one century ago, that this business of firing people to get more money would lead to the crash of capitalism.
Let's see.

[duff_beer_dragon]

The 'Peak Chip' is about size - a threshold has already been reached insofar as how close together the tracks can be packed in any chip ; eventually the tracks e- interfere with one another, and cross-over, and the information is garbled and mixed up.

This is avoidable by making bigger chips, using more than one processor per motherboard / planar board, and - as said above there, changing the manner in which the info. is bussed around..........but, yeah, there is indeed a peak-chip because there is only so far down in scale you can take ICs and processors.

They are adding-machines, so there is a lot of research going on looking into chemical and 'wet' tech. to do the counting ..... I know little about this, don't know what they will use for memory for example, but considering that now we use quartz crystals grown from natural crystal seed, the piezo-electric property being what is required for info. storage, that part probly won't be much of a problem.

( Germanuim chips (wafers) exist too, don't know about any commercial ones? availible? )

[NeoPeasant]

Modern ridiculously overpowered PC's mostly sit around all day converting a hundred or more watts of electricity into waste heat while they wait for the user to send the next keystroke or mouse click.

I expect the power consumption of PC's to become a major issue in the near future when the price of electricity zooms. Raw processor speed will lose emphasis and chipmakers will use MIPS per milliwatt as a selling point.

At home I am using an older slower laptop because the extra speed and power consumption of a fast processor desktop is of no use to me.

Besides I will be able to solar charge the laptop when necessary.

[lotrfanatschool]

GHZ isnt everything. Look at AMD chips, a 2GHZ one can beat a 3.2GHZ P4. This is a good thing, the netburst chips use wayy too much power.

[Falconoffury]

I think AMD is pulling ahead of this chips race with their 64 bit processors. I've had several computers with AMD chips, and they are more cost effective than Intel, and I never had any problems. It's usually the motherboard or some other junk device that breaks down.

[tmazanec1]

Neopeasant (or anybody):
What ARE current PCs (especially laptops) in MIPS per milliwatt?

[bobcousins]

Moore vs Hubbert : it's a showdown!

The underlying picture is slightly more complex here, Intel have been having trouble with their chip architecture for a while. Basically their marketing folks decided the way to beat AMD was with Megahertz, which is/was Intels strength. This means they hit the end stops of transistor size quicker. As has been pointed out, AMD have been doing well with a more efficient architecture.

However, both AMD and Intel, and Moore himself, have said that Moore's law will continue. They have to say that, otherwise a lot of shareholders will get very scared. Moore's Law, if extrapolated, predicts infinite growth. Clearly this can't happen, so a plateau will be reached at some point. We may be hitting a limit with the current technology, we don't know. Even Intel don't know for sure until they try it.

But the wider picture shows the contrast between the "infinite growth" camp and the "peakists". Moore says don't worry, we will use our ingenuity to keep the law going. True, there is plenty of work going on in alternatives, but these are far from production yet.

On the other hand, peakist theory says that growth must be finite.

So this will be an interesting test case to see if the assertion that the market will create alternatives through better technology to maintain continued growth is true, or whether there only so far you can go.

If so, we may also discover how an industry copes with been unable to grow indefinitely, and what effects this may have on the rest of the economy.

Assuming of course, that we can still afford the electricity to run these things...

[chris-h]

$this->bbcode_second_pass_quote('lowem', 'R')euters
- "Peak Chip", anyone? Is this *it* - *gasp* - the End of Moore's Law?

[tmazanec1]

Maybe we will focus on software now...

[pepper2000]

It's what I call the Law of Logical Limits. It says that no trend can continue forever in the physical world. So the question is, what are the limits to growth on processor speed? It seems like they're still a ways off, though.