Need help with Moore's Law!!

[GD695372]

I have a long-term research project due tomorrow on moore's law(the first one), and would appreciate any and all help you could give me. I've been unable, as of yet, to find a formula or anything strictly algebraic. I wouldn't need any help, but I've been extremely busy lately, and have therefore had very little time to research further.

Thanks in advance.

 

[Viper GTS]

A formula? You mean something like:

p(t) = 2(p(t-18))

Where p is processing power & t is time in months.

It's not really suited to be represented as a formula, it's best stated in words. Basically that processing power doubles every 18 months.

Viper GTS

 

[GD695372]

I know, but was hoping that there was a specific formula. Heh a law w/o a formula.

 

[Scrapster]

Viper,

Where's your link?

 

[Viper GTS]

My link? I did that myself...



Viper GTS

 

[GD695372]

Viper's quote is legitimate. One discrepancy is that it doubles every 18 to 24 months though.

Links to any well-written and decipherable papers on Moore's Law would be nice, assuming anyone has any.

 

[Scrapster]

Nevermind. Found one.

 

[GD695372]

Anyone ever done a report on Moore's Law? I'm completely lost at this point.

 

[axelfox]

You need Moore processesing speed in 18 to 24 months

 

[CQuinn]

You can't really apply a formula to Moore's "Law" because it isn't really
a law, or even a theory. It was based on an observation that Gordon
Moore made of how fast he could double the complexity of a chip design
over a given amount of time. Note: This observation was made back around
1965. Most people make the mistake of equating complexity with speed,
and hence the marketing departments of all the computer companies try to
push people into buying "double the speed" every year and a half.

It is more (no pun intended) of a self-fulfilling prophecy than a
scientific principle; since the companies think they can double
complexity every 18 months, they strive harder to make it happen.

In actuality some areas of the industry have exceeded Moore's estimate
in developing new technology, take a look are trends in graphics
and hard drive development for examples.

some links for you.

<a target=new href="http://
http://www.usnews.com/usnews/issue/000710/moore.htm.">Leaving Moore's Law in the dust</a>

<a target=new href="http://
http://www.intel.com/pressroom/archive/speeches/GEM93097.HTM.">&quot;An Update on Moore?s Law&quot;</a>

The rest of the results from a quick Google search

 

[Scrapster]

The whole law looks like BS. Maybe I'm just more used to seeing newton, einstein, and eulers laws that Moore's law sounds like a desperate wanna-be. Do we really need laws for this stuff?

 

[rahvin]

<<It's not really suited to be represented as a formula, it's best stated in words. Basically that processing power doubles every 18 months.>>

As noted above this quote is incorrect, the correct phrase is complexity instead of &quot;processing power&quot;. In specific, in his speach Moore was refering to the number of transistors on the chip. So in effect the real meaning is:

The number of transistors in microprocessors double every 18 months.

This is the reason for the rapid advances in manufacturing technology and why you have seen processors go from .50 to .18 micron technology (in the past 5 years or so), smaller manufacturing ability results in more transistors per square area.

 

[rahvin]

<<The whole law looks like BS. Maybe I'm just more used to seeing newton, einstein, and eulers laws that Moore's law sounds like a desperate wanna-be. Do we really need laws for this stuff?>>

Don't be dumb, it's not a scientific law as noted above, it was an obervation, and it's apparently self fullfilling. It's just called a law because some reporter quoted it as that 50 years ago.

 

[Viper GTS]

Technically it does refer to complexity, or more specifically logic density - how many transistors we're able to pack into a given area. Simply comparing the number of transistors in CPUs is not sufficient, as it doesn't take into account die size.

Viper GTS

 

[Buddha Bart]

this whole thing annoys me, it is not a law
It was an off hand remark that has been jokingly refered to as a law.

bart

 

[Scrapster]

It's just called a law because some reporter quoted it as that 50 years ago.

Exactly. That's what makes it BS.

 

[GD695372]

It's not B.S.. It's a law because Chipzilla says it is. Basically Intel will be able to use the law as a pacecar until either AMD pushes them to do otherwise, or they run into physical limitations of conventional semiconductor technology. I'm assuming that the imaginary barrier is around 0.07micron. I'm also giving AMD a 50:50 chance of beating it.

 

[GD695372]

Here's what I have for my overview so far:

Introduction to Moore's 1st Law, and a brief description(as well as a brief history)

Algebraic formula governing Moore's Law

Prove/disprove Moore's Law, algebraically, with respect to:
-die size
-cpu
-gpu
-processing power
-cpu
-dsp
-gpu
-asic
-acm
-transistor count
-cpu
-gpu
-
-Contrast with the International Technology Roadmap for Semiconductors

Practical applications of Moore's Law

The future, or lack there-of, of Moore's Law
-physical barriers and the possible leaps in technology beyond
-slowdown in miniaturization as semiconductor technology approaches it's limits

Final words and Conclusions

Any suggestions?

 

[rahvin]

<<I'm assuming that the imaginary barrier is around 0.07micron>>

Extreme Ultraviolet lithography is in production testing right now, they will start selling production capable machines for this around the end of this year. Extreme UV litography will exend the limit to around 0.03microns. After that they are gonna have to get real creative because quantum effects start to play a huge role, and quantum effects can be quite nasty, electrons jumping from one transitor to the other and stuff.

 

[KevinH]

It really bugs you guys that much huh? Tsk, tsk. I guess...

 

[GD695372]

That's something that I've been looking at. Unfortunately this is becoming quite a task. Think I should add a section on justifying Moore's Law as a Law?

Include some additional information with respect to rev 1?
http://www.usnews.com/usnews/issue/000710/moore.htm

Anyone know how to model the future of Moore's Law w/o resorting to quantum physics?

 

[rahvin]

The future beyond UV Litho is very fuzzy, I wouldn't mention it more than in passing. Technology moves in mystereous ways.

 

[Capn]

Only way you could prove anything about Moore's &quot;law&quot; would be to do statistical hypothesis testing. I.E. you assume that the exponential growth rate is a certain value perhaps, then you'd have to have data for the actual growth rates over time. This is kind of difficult project, but can be done. Easiest thing to do would be to look at just CPU's because there's a good amount of data. You need a way to quantify performance and this kind of thing may be difficult. There's no way to do it with pure algebraics. Since it is not a law, it is a general trend quoted a long time ago.

 

[beer]

I did an entire project on Moore's Law freshman year. I have tons of MIPS/MFLOPS graphs from 1990-1999 (P3 450 was the top speed at the time). PM me if you want a hookup.

I developed a strictly algabraic formula that you may find useful. Again, PM me.

 

[falconx88]

moores law...

anything that could go wrong will..