Moore's law

[Nipun]

Intel co-founder Gordon E. Moore, who described the trend in his 1965 paper.The paper noted that the number of components in integrated circuits had doubled every year from the invention of the integrated circuit in 1958 until 1965 and predicted that the trend would continue "for at least ten years". His prediction has proven to be uncannily accurate, in part because the law is now used in the semiconductor industry to guide long-term planning and to set targets for research and development

in other words Mr. moore say "over the history of computing hardware, the number of transistors on integrated circuits doubles approximately every two years"

so that the reason still you get newer version and speed up versions of chipsets and processors more often . most of researches want to keep up with Mr. moores for not breaking his law.

On 13 April 2005, Gordon Moore stated in an interview that the law cannot be sustained indefinitely: "It can't continue forever. The nature of exponentials is that you push them out and eventually disaster happens". He also noted that transistors would eventually reach the limits of miniaturization at atomic levels:

In terms of size [of transistors] you can see that we're approaching the size of atoms which is a fundamental barrier, but it'll be two or three generations before we get that far—but that's as far out as we've ever been able to see. We have another 10 to 20 years before we reach a fundamental limit. By then they'll be able to make bigger chips and have transistor budgets in the billions.

In 2003, Intel predicted the end would come between 2013 and 2018 with 16 nanometer manufacturing processes and 5 nanometer gates, due to quantum tunnelling, although others suggested chips could just get bigger, or become layered. In 2008 it was noted that for the last 30 years it has been predicted that Moore's law would last at least another decade.

Some see the limits of the law as being far in the distant future. Lawrence Krauss and Glenn D. Starkman announced an ultimate limit of around 600 years in their paper, based on rigorous estimation of total information-processing capacity of any system in the Universe, which is limited by the Bekenstein bound.

One could also limit the theoretical performance of a rather practical "ultimate laptop" with a mass of one kilogram and a volume of one litre. This is done by considering the speed of light, the quantum scale, the gravitational constant and the Boltzmann constant, giving a performance of 5.4258*10^50 logical operations per second on approximately 10^31 bits






hope this helps people who didn't understand why intel always release new versions again and again and again rather than keep one product for period of times
and hope this helps for everyone's knowledge

[BobF64]

The important part is that Gordon Moore "noted", he didnt define nor require that such things happen, it was purely an observation about how current trends had been going and how they were likely to continue.

Intel, on the other hand, seem to have desperately kept to the pattern mainly to ensure that "Moore's Law" remains valid, rather than producing new chips at a rate suitable for the markets into which they are sold.

[Nipun]

The important part is that Gordon Moore "noted", he didnt define nor require that such things happen, it was purely an observation about how current trends had been going and how they were likely to continue.

Intel, on the other hand, seem to have desperately kept to the pattern mainly to ensure that "Moore's Law" remains valid, rather than producing new chips at a rate suitable for the markets into which they are sold.

definitely

[kalniel]

hope this helps people who didn't understand why intel always release new versions again and again and again rather than keep one product for period of times

Moore's law is descriptive, not proscriptive, so I don't follow your logic in the above statement.

[Nipun]

Moore's law is descriptive, not proscriptive, so I don't follow your logic in the above statement.

no actually what i mean is moore's law is now not very suitable. because these days what intel do is try to keep up with it but not make perfect products for the market. for example. people always dont need a double speed their processor every 18 months.example for me i like the processor get more efficient , include new architectures and new functions rather than speed up always. it will be very usefull for the community

[wasabi]

I think a lot of people have misconception of what Moore's law is. For years people thought it was synonymous with going from a P70 to P133 to P2-350 to P3-700 etc etc

I genuinely think that for most people (i.e. clients), the absolute computing requirement is roughly that of a Core 2 processor from 4 to 5 years ago. Modern computers aren't selling well that are any faster than that, and smartphones will soon, I suspect, stagnate around that level too. Sure, some people (mainly gamers) need more, but not many. Off-machine computing ('cloud' / server) is where the heavy lifting is being done. So progress will be slow until something truly gamechanging happens.

However I think the big architectural changes will be at the server level in client/server model. Much of what I see happening there is a huge increase in virtualisation. Some see GPU/compute taking off too - me I'm not so sure it won't be more hybrid.

[phil4]

Surely what's required is some fancy new feature that needs shed loads of processing and simultaneously is something everyone wants/can use.... I'll be wrong but things I'm thinking of are things like proper AI, and or video/voice recognition, that works.

What I mean by that is imagine an AI instance that acts more like a "companion" with learning, context etc etc.

I'm guessing that'd require some pretty heft processing and perhaps is a feature that may drive that requirement.

[kompukare]

Surely the most point of 'Moore's Law' was that ever 18-24 months the amount of transistors you could economically put in a given wafer area would double?

Well, while nobody really has access to Intel's internal costings the costings which are avail (mainly TSMC) indicate that for example 40nm > 28nm was a lot more expensive per wafer.

So that, yes density doubled but costs also rose so that double transistors for the same cost did not materialize (or not yet anyhow) and there's every indication that 20nm will repeat that. Wasn't that why Nvidia's JHH threatened to throw his rattle out of his pram?

[Pob255]

One thing to note here is that it's transistor count not "speed" or "power" Moore's observation was based on improving manufacturing techniques shrinking the transistors and increasing die sizes.
But as Moore himself and intel have said there is a limiting size at the atomic level which means you cannot go on indefinitely shrinking.

[kompukare]

But as Moore himself and intel have said there is a limiting size at the atomic level which means you cannot go on indefinitely shrinking.

When that limit is reached* things should get interesting. ATM and at least since the mid 1980s Intel has generally been in the lead process wise since they make so much money. So for the last 15-20 years at any rate, they have always been able to afford to have the best fab so that even when their designs were poor (P4) they weren't in that much trouble.

But when the limit is reached (say it's 4nm), I'm sure Intel will be the first but a year or so afterwards TSMC will reach that limit too and eventually any other fab still around will too. At that stage everyone will be on the same node, so Intel's node advantage will disappear. While Intel obviously have a massive R&D budgets for chip designs once there is no more node advantage new designs will no longer be able to utilise an increased transistor budget. Should get very interesting.

*not sure when since 14nm is on Intel's schedule and 10nm and 7nm I think are in the lab