Are Cheap Laptops a Roadblock for Moore's Law?

Timothy Harrington writes "Cnet.co.uk wonders if the $100 laptop could spell the end of Moore's Law: 'Moore's law is great for making tech faster, and for making slower, existing tech cheaper, but when consumers realize their personal lust for faster hardware makes almost zero financial sense, and hurts the environment with greater demands for power, will they start to demand cheaper, more efficient 'third-world' computers that are just as effective?" Will ridiculously cheap laptops wean consumers off ridiculously fast components?"

No

[An Ominous Coward]

Given that Moore's Law is that the number of transistors on a chip doubles every 18 months, no. Even if the gigahertz / number of cores war stops for laptops, there's lots of components that can be put on chip. But apparently it's too much to ask from a rag like CNet to get their basic definitions correct.

Cheap laptops

[backslashdot]

Fact is.. people are installing Vista on laptops that really shouldn't be running them..

People use their cheap underpowered laptop, get frustrated ..curse Dell and Microsoft. And then go to a nicely performing (but more expensive) Mac.

If laptop makers didn't tempt consumers with their underpowered crap, maybe they would have a decent reputation. I don't see how Moore's law is affected.

Apple is the only computer manufacturer whose low end PC's actually perform tolerably.

Machrone's Law

[michaelmalak]

Moore's Law as it applies to PCs has its own "law": Machrone's Law [ieee.org]. It's not as strong a "law" as Moore's as it has had to undergo continual adjustment, but there is a definite phenomenon. Also related is the amusing Wirth's Law, also described in that IEEE Spectrum online blurb.

Jeesh

[I'll Provide The War]

"Moore's Law" has nothing to do with performance.

http://arstechnica.com/articles/paedia/cpu/moore.a rs [arstechnica.com]

Gordon Moore: The complexity for minimum component costs has increased at a rate of roughly a factor of two per year. Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years.

Instead of placing twice as many transistors on a cpu you can instead place twice as many cpus(a few less for the sticklers) of the same transistor count on a single wafer. Even if consumers no longer care about FLOPS they will still be swayed by lower cost, longer battery life, smaller dimensions and passive/quieter cooling.

Re:Of course it won't halt moore's law

[IceCreamGuy]

It was more of a statement pertaining to the cost of transistors, the number of transistors was just one part of the concept. Ars Technica has a great article on it by Jon Stokes: http://arstechnica.com/articles/paedia/cpu/moore.a rs/1 [arstechnica.com]

Re:somebody doesn't understand Moore's Law

[flaming-opus]

These two things are coincident, but not correlated.

multicore is becoming popular because instruction-level-parallelism has approached a practical limit, not capacitance. Basically processor designers are getting all these "free" transistors, and don't know what to do with them except add cores.

Processor speed limits come from heat generated by switching speeds, combined with heat from leakage current. Improved transistor density actually improves the heat generated by switching, but has to be balanced against the increased leakage current from a smaller lithography process.

Misunderstanding Moore's Law?

[argent]

Moore's law is not about exponential increases in absolute performance, it's about exponential increases in performance PER UNIT COST. The original formulation was based on the fact that the number of transistors in a chip using the CHEAPEST transistors was doubling every 24 months.

It doesn't matter whether you get twice the performance for the same price, or the same performance for half the price (and half or less the power usage), you're still following Moore's Law.

The really interesting thing is that Moore's Law applies to everything we make. The doubling time depends on the technology, but the best performance-per-unit-price for every technological product from oxcarts and clay tablets to rockets and ebooks can be shown to follow an exponential curve back as far as we have hard enough figures to plot meaningful points.

Re:Moore's Law in Dynamic Equilibrium?

[cliffski]

people are already slowing down their lust for faster, higher spec devices. Hardly anyone I know has any plans to upgrade their PC at any point in the forseeable future, and having just bought a new battery for this laptop, I'll happily keep using it another 3 or 4 years. The final push which would make me get a new one would be a less weight, longer battery life, or lower power drain. Or maybe a solid state drive or something with no fans. As far as computing power goes, my laptop surfs the web, sends email and plays a few low-sys req games, 99% of its features are unused, especially the CD burner.
Quieter, lighter, and low-power are the new fast.

theoretical vs practical

[perlchild]

And here I thought Moore's Law applied to the top of the line chip designs, from manufacturers, not units sold...

Not that they automatically are incompatible, but Moore's law seemed to pace "research" a lot better than market, ever since I first heard of it...

The low-cost laptop units are among the first units I've seen to approach what customers really want, as opposed to what manufacturers want... Meaning the olpc won't be "necessarily" obsolescent in a year... And even if it was, people would(wisely, I might add) refuse to pay another 100$ next year...
Which isn't to say bundling a low-cost laptop, with say, internet service(as I've heard bandied about) might not work...

Re:somebody doesn't understand Moore's Law

[HardCase]

Processor speed limits come from heat generated by switching speeds, combined with heat from leakage current. Improved transistor density actually improves the heat generated by switching, but has to be balanced against the increased leakage current from a smaller lithography process.

Yes, that's true, but do not discount the effects of die capacitance. Each transistor presents a load to the signal, each interconnect presents a time delay and when you put them together, you have to overcome the problem of the edge rate of the signal either slowing down and failing to meet voltage margins or overshooting the voltage margins and damaging the transistors. The compromise must also meet the logic's setup and hold requirements and the interconnect has to be sufficiently short for the signal to transit in less than one bit period.

Heat is easy to deal with. Bigger heat sinks, more air flow. Moving signals from one side of the die to the other...that's hard. Parasitic capacitance, inductance and resistance make it harder.

Now, if you really want to make it hard, try moving those signals from the CPU to the memory controller, then out to the memory modules. If you thought that getting a high speed digital signal across a centimeter of silicon was hard, try moving it across six inches of PCB.