Talk:Moore's law/Archive 3

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Request for picture

Is there enough images in this article now so that we can remove the "reqphoto" template? Jaqian (talk) 12:26, 6 April 2009 (UTC)

Although the article seems fine to me, the template does ask for a photo rather than a graphic. I wonder if a photo of Moore would be good, especially lecturing, or perhaps a photo of the original magazine. - Wgsimon (talk) 17:29, 6 April 2009 (UTC)

Fair use rationale for Image:GordonMooresOriginalGraphFrom1965.PNG

Image:GordonMooresOriginalGraphFrom1965.PNG is being used on this article. I notice the image page specifies that the image is being used under fair use but there is no explanation or rationale as to why its use in this Wikipedia article constitutes fair use. In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use.

Please go to the image description page and edit it to include a fair use rationale. Using one of the templates at Wikipedia:Fair use rationale guideline is an easy way to insure that your image is in compliance with Wikipedia policy, but remember that you must complete the template. Do not simply insert a blank template on an image page.

If there is other fair use media, consider checking that you have specified the fair use rationale on the other images used on this page. Note that any fair use images uploaded after 4 May, 2006, and lacking such an explanation will be deleted one week after they have been uploaded, as described on criteria for speedy deletion. If you have any questions please ask them at the Media copyright questions page. Thank you.

BetacommandBot 06:54, 7 November 2007 (UTC)

Fixed this (I hope) by using the same templates as Image:Valeri borzov.jpg ---- CharlesGillingham 19:29, 7 November 2007 (UTC)
I guess I failed. I'm not sure why my template was insufficient. One bot marked the image for deletion, another bot deleted it. I'm not sure if the second bot even noticed that I'd tried to fix this. Does anybody understand how this works well enough to try to get the image back? ---- CharlesGillingham 04:40, 4 December 2007 (UTC)

LCD screens do not follow Moore's law

The page currently states that LCD screens also follow Moore's law. This is patently incorrect. In 1997, the standard laptop display had a resolution of between 640x480 and 1024x768 pixels. By 2000, it had increased to between 1920x1440 and 1024x768. Since then, it has decreased: the maximum resolution you're likely to find is 1920x1200, and it's very rare to find anything more on the desktop. If Moore's law had applied over the last 10 years, we would now have displays ranging between 3600x2700 and 5600x4200.

The reason this hasn't happened is clear: the web's holding it back. Most web sites don't even render correctly at 1920x1440, and the tendency seems to be to more breakage. HTML is holding back progress here.

I've removed the reference to LCD screens on the main page. Groogle 00:05, 10 November 2007 (UTC)

It is the market that is holding it back - no one wants to buy hi-res monitors because they don't realize that if they had them they could read 30% faster, just like on printed paper (at 300 dpi). When the public figures this out you will see the standard at about 5600x4200, just like you predicted. 199.125.109.48 (talk) 18:08, 10 February 2008 (UTC)
Wait, what? How would increasing the resolution make reading faster? Lucifer-oxy (talk) 04:44, 12 October 2009 (UTC)
Yeah this is all nonsensical reasoning here. The resolution required by our eyes' retina is oversatisfied as it is. Your eyes won't be capable of seeing the difference reading from a screen with more pixels per mm. What still *does* speed up in screens exponentially is the time it takes to draw pixels and switch them in color and/or status. Just look at the new Samsung screens used in HTC smartphones or the iPhone 4. Not only is the amount of dots per inch still improving, processing graphics is, in contrast to what the TS has written, indeed following Moore's Law as expected. In fact, the bigger driving force behind upping processing power (CPU and GPU) is exactly what people want on screen, which is CLEARLY showing its results. 146.50.227.64 (talk) 07:59, 2 July 2010 (UTC)

Processor Transistor Count, Processor Cache, and Processor Performance

There is a misconception in "Software: breaking the law" section in the article about the relationship between the growth of processor transistor count, and processing power. This section, along with other sections of the article assumes that processing power increases proportionally, specifically at the rate of Moore's Law, to processor transistor count, which is false.

The majority of transistors on a newer processor is allocated to the processor (on-die) cache, in fact processor cache has been known to account for ~45% of a processor's transistor count. The Pentium M Banias has ~77 mil. transistors while Dothan (doubled L2 cache, die-shrinked Banias with minor improvements) has ~140 mil. transistors, however the performance difference between the 2 processors is only around 10-20% (compared to 45% increase in transistors), this proves the disparity between processor power and processor transistor count. There also cases (see benchmarks between Intel Pentium 4 Prescott-2M and Prescott) where an increased amount of cache (more transistor count) has provided negligible or even a decrease in performance (the issue is related to cache size and latency).

Although there is a general trend that increased transistor count of a processor means increased performance, the rate at which both of these factors change are different. Specifically, processor performance have several factors that one needs to consider such as intergrated memory controller, processor microarchitecture (processor effeciency), and clockspeed (the difference of price for processors is usually not the difference of processors themselves, but the clockspeed at which the chip is set to run at, which affects performance, processor stability at high clockspeeds is also an issue), all of which can alter the determination of a processor's power or performance independent of transistor count.

This misconception of processing power and its association with Moore's Law is fairly wide spread (reason below) and should be noted in the article. Part of the reason that it's so wide spread is because when people see the Moore's law graph, they often associate with what they understand (The brand names of the processors i.e.) Pentium 4) rather than what's important (understanding of transistor count). Since the Moore's Law graph shows the chronological order of intel's processor in the correct order (mostly) (i.e.) Pentium 4 is plotted after Pentium 3, which is plotted after Pentium 2, each having more transistors than their predecessors) and people know for a fact that pentium 4 is more powerful than pentium 3, which is more powerful than a pentium 2, the misconception of transistor count and performance is obvious. Also, one has to have a certain degree of knowledge about processors to understand the difference between transistor count and performance, which makes the issue all the more notable. --68.147.51.204 (talk) 08:58, 12 December 2007 (UTC)

I've added a mention about this in the other considerations section of the article. --136.159.209.101 (talk) 18:53, 12 December 2007 (UTC)

I believe the article only claims that all the various measures are increasing exponentially, not that they are increasing at the same exponential rate (or a proportional rate). The article also takes pains to describe the trend as "approximate" and "rough" (in the introduction, anyway).
In some of the other sections, if there are misstatements, you should correct them. I think this is an easy fix. (For example, "Self-fulfilling prophecy" states that the industry aims for "a specified increase in processing power". Should this say "a specified increase in transistor density"? Or are both statements true?) ---- CharlesGillingham (talk) 19:34, 14 December 2007 (UTC)

performance and power

Can we please replaced the word power with performance where the latter is meant. I see a potential confusion with power consumption that is also used in the article. Andries (talk) 10:09, 4 February 2008 (UTC)

Cost per unit area

While cost per transistor is being reduced by shrinking transistor size, the cost per unit area is not necessarily shrinking. That is because each new generation of product adds more and more features and layers into the same square centimeter. By considering how much expensive material is deposited and removed per square centimeter in an Intel 45 nm processor vs. an Intel quarter-micron processor, the point should be clear. The graph added for Economic Impact of Moore's Law should describe this pictorially. Guiding light (talk) 10:19, 24 March 2008 (UTC)

This issue of increasing expense is touched on in the section about industry (A self fulfilling prophecy), and could be covered in more detail there. There could also be a new section that details various aspects of digital technology that are not improving exponentially. If such a section was started, in should subsume the section on software. ---- CharlesGillingham (talk) 15:29, 24 March 2008 (UTC)

Memristor

There may be enough published information on the memristor now to add a note about it to the "Future trends" section. (Although our article Memristor IMHO needs some significant improvement.) -- 201.37.229.117 (talk) 17:28, 1 May 2008 (UTC)

Switching energy

It would nice to mention/link in how the switching energy for a single transistor has decreased over time. Couldn't find the appropriate page on Wikipedia. Does it exist?--Michael C. Price talk 20:24, 8 May 2008 (UTC)

Requested move

Moore's Law → Moore's law — Unlike all the others in List of scientific laws named after people, this one has nonstandard capitalization of "Law"; let's fix it. —Dicklyon (talk) 06:12, 28 May 2008 (UTC)

Survey

Feel free to state your position on the renaming proposal by beginning a new line in this section with *'''Support''' or *'''Oppose''', then sign your comment with ~~~~. Since polling is not a substitute for discussion, please explain your reasons, taking into account Wikipedia's naming conventions.
  • Support – of course, as I proposed it. There's no reason this law in particular should be an exception to the usual wikipedia naming convention, which all the other eponymous laws follow. Dicklyon (talk) 04:19, 29 May 2008 (UTC)
  • Neutral - personally I think that the capitalized version is more appropriate, so if this is left capitalized, we should rename the other "laws" to be capitalized. 70.55.85.131 (talk) 04:14, 30 May 2008 (UTC)
The convention at WP:NAME is "Do not capitalize second and subsequent words unless the title is almost always capitalized in English (for example, proper names)." Moore's law is found in books not capitalized nearly half the time, so that's way off from "almost always" capitalized. I think this articulated convention should trump the common notion of what's "appropriate", in the wikipedia context. Dicklyon (talk) 04:47, 30 May 2008 (UTC)
And I don't understand the penchant of the semi-literate to capitalize important words in English. Gordon Moore is a great guy, but I don't see why we should treat his law special in wikipedia, compared to all the others, especially, when there's a clear guideline about it. Dicklyon (talk) 15:43, 30 May 2008 (UTC)
Capitalisation is not a sign of semi-literacy, as you should know.--Michael C. Price talk 18:01, 30 May 2008 (UTC)
No, but over-capitalization sometimes is. Dicklyon (talk) 18:15, 30 May 2008 (UTC)
Well, that's true by definition, but the question is: is it over-capitalisation?--Michael C. Price talk 22:49, 30 May 2008 (UTC)
Over-capitalization is widespread problem in wikipedia that I fight all the time. This is a mild case, and I don't mean to conclude that the guy who did it is less than literate; he was simply not aware of the guideline, most likely. Dicklyon (talk) 22:58, 30 May 2008 (UTC)
  • Comment - I think it should be noted that this article previously used the lowercase version. It was moved in 2006 to the uppercase version by Wikidrone with the comment: "moved Talk:Moore's law to Talk:Moore's Law: both words should be capitalized, it is a specific law (not a general one). Kapische?". swaq 18:46, 30 May 2008 (UTC)
  • Support. The frequency of use elsewhere is irrelevant: our standard convention for So-and-so's theorem, Whatsit's principle, and Whoever's law is to leave the words other than the name uncapitalized. I see no special circumstances that would distinguish this one from the others. —David Eppstein (talk) 22:03, 30 May 2008 (UTC)
  • Support. Why make an exception when we already have established standards that other articles follow? ~Amatulić (talk) 22:25, 30 May 2008 (UTC)
Because the guidelines says we should also be guided by common English usage; Google it and you find about 90% occurrence of the capitalised form.--Michael C. Price talk 23:02, 30 May 2008 (UTC)
If you do google book search, and restrict yourself to non-title cases (since many pubs have different title conventions), the usage is closer to 50%. And the guideline is clear what to do even if it's a majority: "Do not capitalize second and subsequent words unless the title is almost always capitalized in English (for example, proper names)."

Discussion

Any additional comments:

Looks like an admin decided, and did the move. Dicklyon (talk) 05:51, 31 May 2008 (UTC)

When will we reach a xeraflop?

Assuming Moore's law holds and computational power doubles every 18 months in the year 2068 we will break the xeraflop barrier.

Note: to put it into perspective I've listed the equivalent number of petaflops. A xeraflop is over a trillion petaflops!

2008 – 1 petaflop

2011 – 4 petaflops

2014 – 16 petaflops

2017 – 64 petaflops

2020 – 256 petaflops

2023 – 1,024 petafops (1 exaflop)

2026 – 4,096 petaflops

2029 – 16,384 petaflops

2032 – 65,536 petaflops

2035 – 262,144 petaflops

2038 – 1,048,576 petaflops (1 zettaflop)

2041 - 4,194,304 petaflops

2044 – 16,777,216 petaflops

2047 – 67,108, 864 petaflops

2050 – 268,435,456 petaflops

2053 – 1,073,741,824 petaflops (1 yottaflop)

2056 – 4,294,967,296 petaflops

2059 – 17,179,869,184 petaflops

2062 – 68,719,476,736 petaflops

2065 – 274,877,906,944 petaflops

2068 – 1,099,511,627,776 petaflops (1 xeraflop)

Lordvolton (talk) 18:04, 15 June 2008 (UTC)

Lead diagram text

The text for the main diagram says that "As of June 2008 Intel's processors contain 47 million transistors" but without telling us anything about which chip this refers to this is meaningless; also seems rather lower tnan the trend, isn't it?--Michael C. Price talk 21:39, 12 July 2008 (UTC)

While I'm sure Intel still has something kicking around with around 50 million transistors, the new(ish) Core 2 Duo Wolfdale is about 400 million[1]. The Core 2 Quads are even higher. This text was introduced by an anon who first put the number at 4.7 billion (which even Itanium doesn't touch) and then "corrected" it to 47 million. I have deleted the text. — Aluvus t/c 23:51, 12 July 2008 (UTC)
Hi Aluvus, thanks for the clarification. It would be really nice if someone with the latest data could update the graph, which starting to look a bit dated.--Michael C. Price talk 10:10, 13 July 2008 (UTC)

Moore's law and Oil prices

Recently realized that many materials used by industries to advance Moore's law are hydrocarbons or hydrocarbon derivatives, meaning they ultimately derive from petroleum. It makes sense since the relevant polymers (including plastics and solvents) all have a backbone consisting of a chain of carbons, each of which is also attached to at least one hydrogen atom. For certain, key ingredients of photoresists, which are used to shrink features, still continue to use hydrocarbons as feedstock. The petroleum basis of volume production of hydrocarbons makes the connection with oil pricing. So my latest edit reflects this realization. The oil price data is clearly referenced. The hydrocarbon basis of industrial chemicals is well known to the organic chemists. Someone also added a patent reference, which is a good example of the photoresist, but since petroleum is clearly used for more needs than just gasoline, it is clear that its impact would be widespread. I hope this also calls more attention to the need for finding alternatives to petroleum as a source for these needs.Guiding light (talk) 01:35, 11 August 2008 (UTC)

Whose realization are we talking about here? If your own, then it's WP:OR, and not an appropriate thing for wikipedia. In order to make a connetion of these well known things (increasing oil price and use of hydrocarbons in organic chemical industries) to Moore's law, we really need a source that makes that connection. It really dilutes the point of Moore's law to call a ragged turn upward in oil prices an "exponential"; Moore's law is pretty accurately an exponential over many orders of magnitude; oil prices are not. I don't deny the point that rising petroleum prices will affect many industries, as for example mentioned in a letter to editor of the SF Chronicle today (written by my wife). But inferring a connection to Moore's law still needs a source. Dicklyon (talk) 03:31, 11 August 2008 (UTC)
My point of including the oil price curve is to add a tangible example of the economic barrier. Actually I hardly need to make the connection through the hydrocarbons that are used - it is only because I am familiar that I felt it was particularly relevant. The oil price connection is clear enough by considering the general impact on the economy (for example, the cost of gasoline used in the delivery of products exemplifying Moore's Law). Since the connection is obvious, what is the need for a source? You just need to see how much oil prices have risen.Guiding light (talk) 04:18, 11 August 2008 (UTC)
I did some more searching for an adequate source. It wasn't hard to find (I knew it was an obvious issue). So I added them to the edit I had earlier. One of the references added (the C&EN article) will address the relevance of oil prices to semiconductor manufacturing (it is explicitly stated there:"Everything in photoresists comes from crude oil") and the rising costs of semiconductor manufacturing are already understood to be relevant to Moore's law. Guiding light (talk) 05:17, 11 August 2008 (UTC)
Just because you see it as obvious doesn't mean it's verifiable (per WP:V). The source of the figure you added does not appear to support the assertion in the caption, "The price of crude oil is increasing exponentially." And the rising costs of materials has not been connected to a possible limit to Moore's law, as far as I know, and certainly not in the source you cite. For now, I'll leave it for others to decide about, as I've already reverted it twice. Dicklyon (talk) 05:40, 11 August 2008 (UTC)
Compromise. I agree with Dicklyon that we haven't found a reliable source that claims that the price of polymers & photoresists is increasing (even vaguely) exponentially. I agree with Guiding Light that he has found reliable sources that shows that they are increasing (just not exponentially), and that this effects chip manufacture. I've edited the article to remove the word "exponential" from the description of the trend. ---- CharlesGillingham (talk) 07:51, 11 August 2008 (UTC)
I notice User:Dicklyon has removed the section again. I went back and read through the sources for the section and I think I now agree with him that we don't have a source that specifically says "rising petroleum is one of the most important reasons that it is more expensive to build chips." We need a source that says exactly this, otherwise we have a problem of WP:UNDUE weight. Ideally we need a source that says: "the things that make it hard to keep up with Moore's law are (a), (b) and (c), in that order." I think the Infoworld article comes close to saying this, although I'm not expert enough about the chip industry to judge. Could we use the infoworld article as a source for the first paragraph? I'm posting the link here so that a more expert editor can use it: 2005 Infoworld article on Moore's law impact from rising costs and diminishing returns. ---- CharlesGillingham (talk) 19:53, 23 August 2008 (UTC)
Well as this section stands there are no citations whatsoever. Clearly industry is facing increasing cost barriers to sustaining Moore's Law. The Moore's Law trend up to now is not being denied but clearly any trend has a cost to be supported. Now, as long as weight is given to semiconductors and chip manufacturing as key to the trend, we need to keep the original cited statements and possibly add more. At the very least we need to draw up the list of items such as materials prices, mask tapeout costs, new processes, more layers being stacked in chips, etc. The first two items I have already cited, I can find some for the next two. These four items already comrpise all or most of the cost issues.Guiding light (talk) 01:08, 24 August 2008 (UTC)
I have no objection to the idea that rising costs may slow Moore's law. My objection is to expanding and expounding on this idea without sources. The rest of the section, as you note is all unsourced, and I will remove the whole thing if this is not rectified within a few weeks. The issue is not about finding sources that prices are rising, but rather sources that say something about the connection of that to Moore's law. If you have sources, by all means write about what they say. But WP:NOR please. Dicklyon (talk) 01:28, 24 August 2008 (UTC)

I'm glad to see some sources here, but I'm a little concerned that the whole section is no longer about an important topic. This section was about how the semiconductor industry uses Moore's law for planning. That's an interesting and important subject, which I think any article on Moore's law should cover. Now the section is only about one tiny-tiny factor in a little aspect of chip manufacture tangentially related to Moore's law. Can't we find sources that describe how Moore's law is used? Do any of these new sources do that? ---- CharlesGillingham (talk) 06:37, 24 August 2008 (UTC)

I found sources for the old first paragraph, which I think is a very important point. It took me about twenty minutes. ---- CharlesGillingham (talk) 08:06, 24 August 2008 (UTC)

Sorry when I was working on this section, I took out pretty much everything, with the intention of bringing stuff back once I could find the sources. I think splitting off the manufacturing and the self-fulfilling aspects is a good idea. Guiding light (talk) 08:25, 24 August 2008 (UTC)
Ah, okay. Sorry for jumping the gun (and for smart remarks). ---- CharlesGillingham (talk) 08:32, 24 August 2008 (UTC)

Updated image source

Hi there,

As I have written here before more then a year ago, and others complained about it aswell, I took the liberty to search for more accurate statistics about the current trent in transistor count. So instead of the old image used here that goes to 2004 (Itanium chips) this one goes even into the future:

http://www.indymedia.org.uk/en/2006/05/340331.html

So is this something for inclusion? What do you think? :)

22:31, 12 August 2008 (UTC)

It's good to have a graph that is updated for processors up to now. I was also thinking, now that Intel is focusing more on Atom processors, wouldn't this be a diversion from the curve. The Atom processor contains less than 100 million transistors, so it is a significant step backwards.Guiding light (talk) 08:38, 24 August 2008 (UTC)
Hardly a step back, unless that is their leading chip. --Michael C. Price talk 08:49, 24 August 2008 (UTC)
At least not yet. It should belong in another category, like consumer electronics. Yet it is being used to substitute standard Core architecture chips in low-end notebooks like eeePC.Guiding light (talk) 12:09, 24 August 2008 (UTC)
I was just reading the second reference (Moore's 1965 article "Cramming more components onto integrated circuits") and it looks like a processor like Atom qualifies as the integrated circuit containing "minimum cost components." Likely more so than the larger die, usual Intel processors. Larger dies are harder to yield due to their loss at the wafer edge. The number of components in the popular Atom processor has fallen back to 47 million, from more than a billion for Itanium or nearly a billion for a Core 2 Quad. Mainstream Core 2 Duo processors are at least 200 million transistors. So that does imply a breaking from Moore's law by Intel itself.Naturalion (talk) 06:01, 25 August 2008 (UTC)
If cost is entering the picture I suggest we plot it on a super-Moore's law graph. --Michael C. Price talk 08:31, 25 August 2008 (UTC)

GA Reassessment

This discussion is transcluded from Talk:Moore's law/GA1. The edit link for this section can be used to add comments to the reassessment. In order to uphold the quality of Wikipedia:Good articles, all articles listed as Good articles are being reviewed against the GA criteria as part of the GA project quality task force. While all the hard work that has gone into this article is appreciated, unfortunately, as of August 23, 2008, this article fails to satisfy the criteria, as detailed below. For that reason, the article has been delisted from WP:GA. However, if improvements are made bringing the article up to standards, the article may be nominated at WP:GAN. If you feel this decision has been made in error, you may seek remediation at WP:GAR.

  • I count eight requests for citation in this article, some of them outstanding since August 2007. All citation requests resolved. --Kvng (talk) 21:15, 27 November 2011 (UTC)
  • A self-fulfilling prophecy: industry struggles to keep up with Moore's law has been tagged as needing improvement since February 2008. Resolved by previous edits. --Kvng (talk) 21:17, 27 November 2011 (UTC)
  • The Software: breaking the law section is entirely unsourced. Section no longer exists. --Kvng (talk) 21:17, 27 November 2011 (UTC)
  • All of the web sites relied upon as sources need to have last access date and publisher information. Fixed. --Kvng (talk) 23:17, 27 November 2011 (UTC)
  • There are two dead links.[2] Fixed. --Kvng (talk) 23:17, 27 November 2011 (UTC)

--Malleus Fatuorum (talk) 01:38, 23 August 2008 (UTC)

Lost text

This material was lost in a recent edit-skirmish, I'm assuming because it was unsourced. These are, I think, important points. We have new sources for this section that verify both of these points. Could someone who has read the sources and connect these points to them? ---- CharlesGillingham (talk) 18:34, 25 August 2008 (UTC)

I added the graph for the second item. The mask cost has been put back after source found. I didn't add the design starts part, but it is a reasonable point. I am looking for sources for these points.Guiding light (talk) 22:43, 25 August 2008 (UTC)
Economic impact from rising manufacturing costs
1 The time pressure to remain competitive
A typical major design project (such as an all-new CPU or hard drive) takes between two and five years to reach production-ready status. In consequence, component manufacturers face enormous timescale pressures—just a few weeks of delay in a major project can spell the difference between great success and massive losses, even bankruptcy. Moore's law suggests phenomenal progress for technology over the span of a few years. Expressed on a shorter timescale, however, this equates to an average performance improvement in the industry as a whole of close to 1% per week. Thus, for a manufacturer in the competitive CPU market, a new product that is expected to take three years to develop and turns out just three or four months late is 10 to 15% slower, bulkier, or lower in capacity than the directly competing products, and is close to unsellable.
2 Falling profit margins
As the cost of semiconductor equipment is expected to continue increasing, manufacturers must sell larger and larger quantities of chips to remain profitable. (The cost to tape-out a chip at 180 nm was roughly US$300,000. The cost to tape-out a chip at 90 nm exceeds US$750,000, and is expected to exceed US$1,000,000 for 65 nm.[citation needed]) Despite the increasing costs, the final product retail price cannot be expected to increase; generally, they tend to stay within the same price range. As a result, product margins are expected to follow a decreasing trend. In recent years, analysts have observed a decline in the number of "design starts" at advanced process nodes (130 nm and below for 2007). While these observations were made in the period after the 2000 economic downturn, the decline may be evidence that traditional manufacturers in the long-term global market cannot economically sustain Moore's law.
Yes such essays definitely need to be backed by sources. Dicklyon (talk) 18:02, 5 September 2008 (UTC)

Please improve graph "NAND_Flash_accelerates_Moore's_Law.JPG"

First of all -- kudos to all who have worked on this article. I think that it is overall very good, including the graphs. However, I do have a request for clarification in the graph "NAND Flash accelerates Moore's Law" ( NAND_Flash_accelerates_Moore's_Law.JPG ). It's very difficult for a lay person to understand what this is representing.

(a) The other graphs in this article show "improvement" as increasing slope, this shows improvement as decreasing slope.

(b) It supposedly represents "acceleration", but shows straight lines on linear scale -- where's the acceleration?

(c) It seems to me that this would be improved by including for comparison a "Moore baseline" of (or "corresponding to") "doubling every 24 months".

Obviously I'm missing the technical points in this, but that's exactly my point.
Can we change this graph -- or at least its accompanying caption -- to clarify matters?
Thanks. -- 201.17.36.246 (talk) 14:34, 5 September 2008 (UTC)

I added to the graph caption to make things clearer.Guiding light (talk) 08:58, 18 September 2008 (UTC)
I took out the graph and the statement, as it appears to be WP:OR. The idea that you can look at design rules in one product category and extrapolate to an accelaration of Moore's law seems to betray a fundamental misunderstanding; still if there's a source for this claim, we could include it and attribute it appropriately. As for the figure, we at least need a source for the data if we're going to include it. Dicklyon (talk) 02:06, 9 October 2008 (UTC)
The NAND trend has been updated in a new graph with reference. The doubling of complexity (number of components per unit area) is faster than every 18 months. I think the confusion comes from the insistence on only being one Moore's law. Clearly this technology is accelerating faster than the original Moore's law. By itself that trend that would make Moore's law outdated. But if we include it as a supporting evidence for the Moore's law trend, you would have to say that it is an accelerating force. Just as we may expect decelerating forces such as economics, materials costs, etc.Guiding light (talk) 15:39, 23 March 2010 (UTC)
The revised graph has pretty severe flaws. It contains only 3 years of actual data, and shows only process nodes without providing any context for their impact on cost or transistor density (that is, the things Moore's Law actually focuses on). Release dates are mixed without comment with announcement dates (and since announcement dates will precede actual release, this artificially increases the apparent pace of change). The presentation is unnecessarily confusing; despite containing only 3 years of actual data, the x axis covers 7 years. But what it boils down to is that you are making a fairly bold claim (if transistors used in NAND flash were shrinking dramatically faster than transistors used in processors, that would be quite notable), without identifying any technical cause, on the basis of 2 node transitions that have actually happened. You have not provided enough data to support the claim that you are making. — Aluvus t/c 03:38, 24 March 2010 (UTC)
The revised graph is becoming really old by now. I mean, Core i7 appeared in Q4 of 2008. That's 4+ threads more than the 4 core 4 threads of the maximum (QX9770) in Q1 of 2008. Seems to me Moore's law is becoming a modest prediction, rather than actual. 146.50.227.64 (talk) 07:48, 2 July 2010 (UTC)

Analog ICs

Just a point I'd like to make is that Moore's law also applies to analog ICs. Analog circuits can more closely approach their ideal behavior when more transistors are available on a chip.--SkiDragon (talk) 03:15, 3 October 2008 (UTC)

Unsourced junk

There has been so much written about Moore's law that it's hard to imagine any excuse for editors adding their own unsourced essays of analyses and futures and such. I just removed a bunch of such junk. Most of the article is well sourced, and a bit more pruning might make it really conform to WP:V. Dicklyon (talk) 16:31, 21 November 2008 (UTC)

Exponential?

Someone went through the article and removed several uses of the word "exponential". Can anyone explain to me why this was done? ---- CharlesGillingham (talk) 08:52, 30 December 2008 (UTC)

Seems wrong to remove them, I've put them back. Wgsimon (talk) 11:04, 30 December 2008 (UTC)

Is Moore's Law Really a Scientific Law?

I believe there should be a section in the article about how Moore's Law is not, by definition, a law. There are many scientific laws, gravity being one of them. The most fundamental characteristic of gravity (and the reason that we call it a law) is that it is an absolute. Here are some of the characteristics of a physical law that Moore's Law does not adhere to:

Absolute. Nothing in the universe appears to affect them. (Davies, 1992:82) Stable. Unchanged since first discovered (although they may have been shown to be approximations of more accurate laws—see "Laws as approximations" below), Omnipotent. Everything in the universe apparently must comply with them (according to observations). (Davies, 1992:83) see the wikipedia entry for "Physical Law"

The problem with Moore's Law is that it is dependent upon humanity. It is intuitively true that the exponential growth of technology could come to a precipitous halt if any number of disasters struck humanity. Gravity will continue regardless of whether or not the human species is wiped out. This is why gravity is a scientific law. Moore's Law is not a law at all, at least not in the proper sense of the the term. If you disagree with me, I welcome discussion. But please do not delete this post simply because you disagree.--Jchrom3 (talk) 00:08, 25 January 2009 (UTC)

It's obviously not a natural law; is there something in the article that suggests it is? If so, let's fix that. Dicklyon (talk) 05:55, 25 January 2009 (UTC)
I can't see how anyone would think Moore's Law is a physical law. There are many kinds of laws. Moore's law is an economic observation, as is Engel's Law, Gresham's Law, Mill's Law of supply and demand and many others. We call it a "law" because of its precise mathematical form. This doesn't mean it's a physical law.
In my opinion, this is obvious and there is no reason to discuss the issue in the article. ---- CharlesGillingham (talk) 15:59, 25 January 2009 (UTC)
It is scientific law in regards to the observation of the development of an industry, like the law of diminishing returns. Although it can be considered derived from physical law, it is not part of physical law, but an observation on the efficiencies of design in the computer industry. Excellent question. —Preceding unsigned comment added by 69.232.205.64 (talk) 10:00, 1 March 2011 (UTC)
Look at the following statement. [3]

So the "best

price/mip trend" is really a 38 year trend of doubling every 1.5 years.


Predicting the future is always hard. There are reasons to expect

things to progress faster (current chips are only 2D planes of transistors - once they go 3D doubling times will be shorter (transistors in 1 cubic-centimeter goes up with 3rd power of size while transistors on 1 sq-centimeter goes up with 2nd power of size)). The doubling time could stay the same (it is a 23 or 38 year trend). Or the doubling time could slow down (as chip features get very small it gets harder to cut the size by 30%). All in all, I think the best bet is that the doubling in 1.5 years rule will keep on working.

Why is this here?

"Extrapolation partly based on Moore's law has led futurists such as Vernor Vinge, Bruce Sterling, and Ray Kurzweil to speculate about a technological singularity. Kurzweil projects that a continuation of Moore's law until 2019 will result in transistor features just a few atoms in width. Although this means that the strategy of ever finer photolithography will have run its course, he speculates that this does not mean the end of Moore's law:



Moore's law of Integrated Circuits was not the first, but the fifth paradigm to forecast accelerating price-performance ratios. Computing devices have been consistently multiplying in power (per unit of time) from the mechanical calculating devices used in the 1890 U.S. Census, to [Newman's] relay-based "[Heath] Robinson" machine that cracked the Nazi [Lorenz cipher], to the CBS vacuum tube computer that predicted the election of Eisenhower, to the transistor-based machines used in the first space launches, to the integrated-circuit-based personal computer.[43]

Thus, Kurzweil speculates that it is likely that some new type of technology (possibly optical or quantum computers) will replace current integrated-circuit technology, and that Moore's Law will hold true long after 2020. He believes that the exponential growth of Moore's law will continue beyond the use of integrated circuits into technologies that will lead to the technological singularity. The Law of Accelerating Returns described by Ray Kurzweil has in many ways altered the public's perception of Moore's Law. It is a common (but mistaken) belief that Moore's Law makes predictions regarding all forms of technology, when it actually only concerns semiconductor circuits. Many futurists still use the term "Moore's law" in this broader sense to describe ideas like those put forth by Kurzweil."

this is just alot of unsourced gibberish. This isn't the singularity page people, and I was so hopeful that we could keep the tinfoil-hat crowd away from this article. I move that this entire section be excised from the article. There isn't one reference, and it just reeks of original research. —Preceding unsigned comment added by 199.17.64.1 (talk) 00:32, 5 February 2009 (UTC)

I don't care if this article covers this topic or not, but I can help out with the source. This section is not original research. It is drawn from Kurzweil's 2005 book, The Singularity is Near, as well as other publications (such as New Scientist) who have covered this idea. I've dropped in this main source right after the topic sentence. Would anyone who has this book mind slipping in a few page numbers? I am also equally happy if everyone thinks we should cut down the section, or cut it all together. ---- CharlesGillingham (talk) 11:40, 6 February 2009 (UTC)
Okay, I fixed the topic sentence, and added Moore's quip as a kind of rebuttal to Kurzweil. I think an educated reader can make up their own mind about whether this is nonsense or not. ---- CharlesGillingham (talk) 12:03, 6 February 2009 (UTC)

What POWERS Moore's Law - the speed of progress, or science-throttling?

Anyone know what exactly powers Moore's Law? (ie, does the law reflect the max speed of scientific progress, or the max speed that companies are willing to improve CPU's each year?) The more I look at it, the more strange it seems, so I just had to do some research. According to that chart on the main page, transistor counts have been RIGHT on target since 1970.

So, my question is: how is this speed maintained so precisely? Is the progression of transistor counts powered by the ACTUAL, NATURAL speed of technological discovery (pushing forward as fast as we can go), or can we go faster, but the pace is being purposely throttled down to a precise speed that is seen as the most profitable?

I did tons of searching all over the net and forums, and couldn't dig up anything on this exact question. I find it very strange that this article doesn't mention what exactly POWERS Moore's Law.--68.111.167.64 (talk) 13:55, 27 April 2009 (UTC)

Read the section "target for industry". This helps explain why Moore's prediction has been so precise -- it determines R&D decisions and planning in the industry.
However, while this describes the precision of the law, it doesn't explain the exponential form of the law. Why is exponential improvement possible? Why is the industry able to keep up? (We know that they want to, but it is less clear why they are able to.) Also, why hasn't improvement been faster, i.e. why hasn't there been some breakthrough that bucked the trend? A subtler version of this question is this: why don't other fields improve exponentially? For example, transportation technology stopped improving exponentially around 1970, after ~150 years of exponential improvement. Why did it stop? Why do some technologies improve exponentially, while others do not?
These questions go beyond economics and get into the history & philosophy of technology. In my view, they do not have an easy answer. ---- CharlesGillingham (talk) 23:27, 27 April 2009 (UTC)

Old sections to archive

Please delete old sections from this discussion page. Most are from 2005/2006...

The way we do this in Wikipedia is to move old comments into "archives". (It's hard to find, but up near the top of this page is a little box called an "archive" box. This provides a link to old discussions.) I've archived the comments before 2007. ---- CharlesGillingham (talk) 02:31, 28 April 2009 (UTC)
Talk:Moore's law/Archive 2 now contains comments from 2007. (I left behind the discussion about the image we lost, because I thought that was a great illustration for this article and I still hope we can get it back some day.) ---- CharlesGillingham (talk) 02:45, 28 April 2009 (UTC)

Data sources

The first link doesn't work.--70.235.84.118 (talk) 13:38, 25 May 2009 (UTC)

The first link under "Data" to wi-fizzle.com works fine for me. So either it was a temporary problem, or you need to be more specific. — Aluvus t/c 18:30, 25 May 2009 (UTC)

Implications of

would make a very interesting sub section. Please consider if you're in the know. (It will be tricky to write considering all wikipedias rules, perhaps it would be easiest to just add relevant links) --193.166.137.75 (talk) 12:42, 3 July 2009 (UTC)

It shouldn't be at all tricky; just find a source or two and report what they say. Start here and here and here. Dicklyon (talk) 15:29, 3 July 2009 (UTC)

Ultimate limits of the law

The first paragraph of the Ultimate limits of the law look like OR to me, and Josephson junctions and quantum dots provide counter-examples to the notion that things can't be controlled down below the quantum scale; the wavefunction is completely controllable and deterministic. It's only the introduction of entropy that cause uncontrollable probablilistic behaviour, and that doesn't have to happen down at the microscale.--Michael C. Price talk 20:31, 30 July 2009 (UTC)

the principle of heisenberg uncertainty at its purest states that you if you wanted to know particle position you could not know its direction (hence velocity), or vice versa, this is because when you try to measure the position of any particle (not just an electron) you disturb its velocity , so you cannot know both velocity and position simulataneously.
http://www.youtube.com/watch?v=TcmGYe39XG0&feature=related this in an informative video from the idian uni of tech IIT Madras i just found on youtube.
additionally http://rationalwiki.com/wiki/Moore%27s_Law this physics site describes the phenomenon in simpler terms, and http://news.cnet.com/2100-1008-5112061.html this site is an article by intel engineers theorizing how soon we might hit this limit, note that if moores law continues at this current pace it will be 2 decades, this is a well understood and uncontroversial principle. from that point to further increase cpu strength one would have to resort to quantum principles , see the article on quantum computing for a more complete explanation. additionally one could physically layer semiconductor chips on top of each other to increase their collective performance, but most of the industry is attempting to harness quantum principles of the standard model.
read the article on quantum computing it will explain everthing, also check citations if youd like to learn in greater depth —Preceding unsigned comment added by 122.106.173.173 (talk) 09:28, 31 July 2009 (UTC)
I shoulld have thought that my first post makes it clear that I'm not questioning the basics of QM (I was a physicist) but your interpretation of its application. You haven't explained how Josephson junctions could work, if nothing is controllable on a microscale.
And please sign your posts and play some attention to formatting. Being an IP address is no excuse for being sloppy.
PS I removed the stuff about supersymmetry.--Michael C. Price talk 10:02, 31 July 2009 (UTC)
The points you make are already made further down in the article (where the ref you cite appears). I put in a link about quantum tunneling for those that need more details, but I don't think we need a detailed physics explanation here, and certainly not one that involves supersymmetry. --Michael C. Price talk 10:15, 31 July 2009 (UTC)

Business practice?

The current version of the article states repeatedly that Moore's law is "effectively a business practice". It seems to me that Moore's law is more than just a business practice. The key prediction of Moore's law is that it will continue to be technologically possible for humans to double the number of transistors on a chip every 18-24 months. This prediction is not at all obvious and it is not as if it's just some arbitrary business schedule that some managers decided upon. It would be nice to have a "business practice" where car manufacturers double their fuel efficiency every two years, but it just has not been possible.--Singularitarian (talk) 10:53, 14 September 2009 (UTC)

Agreed. --Michael C. Price talk 11:32, 14 September 2009 (UTC)
This point is arguable, so Wikipedia should mention both sides of the argument. I moved the "business practice" statement to the end of the lead (some days ago, although another editor and I seem to disagree about its priority. As of this writing, it is at the bottom of the lead.) If you two feel strongly, we could add a sentence like "However, Ray Kurzweil argues that the cause of the exponential trend ... etc." This sentence should summarize the point of view covered in the section of this article dedicated to Kurzweil's ideas. Tread lightly, however: note that some editors feel strongly that this is not worth covering at all (see #Why is this here?, above). ---- CharlesGillingham (talk) 18:03, 14 September 2009 (UTC)
It has returned to the lead, so let's discuss this. Next section. ---- CharlesGillingham (talk)

Defining Moore's law

The key problem I have with the current version of the article is that it mentions a controversial issue in the first paragraph of the article without any rebuttal or source. This seems to me to be WP:POV pushing. As I see it, there are several issues.

(1) Is Moore's law (partly) a business practice? The answer is, of course, yes, and a precise description of the practice is given in the third paragraph of the lead as well as the section about industry in the body of the article.
(2) Do we need to mention this before anything else? I don't think so. We have to assume that some readers have no idea what Moore's law is, and so the first paragraph of the article should simply define it and quit. I think "historical trend" works best, since this is perfectly accurate. If we're going to get into natural-law vs. business practice vs. various theories of scientific progress or social progress, we need sources from both sides and a paragraph or two that outlines the issues. This doesn't belong in the lead.
(3) Is Moore's law a natural law? Obviously not, in my view (see this previous discussion), so obvious in fact that I believe it is hardly worth mentioning. Does the article on Engel's law or the law of diminishing returns make the point that they are not natural laws? However, some readers are unclear on this, so I would prefer to handle this with a single sentence at the end of the lead.
(4) Do we need to mention this first? No, similarly to the point above.
(5) Is Moore's law "just" a business practice? I.e., are there are other factors at work that make Moore's law possible? This is controversial and is the subject of this discussion. Ray Kurzweil believes that the underlying cause is his law of accelerating returns. There are other theories (a brief previous discussion is here). As I said above, Wikipedia must mention both sides of this issue, if it mentions one or the other.
(6) Do we need to mention this first? No, similarly to the other points.
That's my view of the issues, anyway. I'm going to wait a few days before I edit the article. Without further discussion, I will restore it to something similar to this version ---- CharlesGillingham (talk) 19:26, 16 September 2009 (UTC)
Ummmmmmmmm, I'm not sure why you find this worth digging into (no offense intended in the slightest, but I'm not even sure if you're seriously finding fault with the lines being addressed, or are just nitpicking the article's aesthetics... I'm serious, lol). Okay, so:
The line I added has an ENORMOUS informational payload, and delivers right off the bat - within the first words of the article - more than half of the ENTIRE essence of Moore's Law: the fact that it is a business practice more than a natural law (making it even more important to mention PROMPTLY, since the whole article is about something called a "Law" - When I first saw the article, I, as a reader, felt this massive void in the first lines in the lead, which was why I added that in the first place). Everything after that is just details that go on about how the law is used. I see the current lead paragraph as *perfect*, and like the structure of what follows as well. (I really, really don't know why you see the need to discuss anything here) The other line you mentioned, at the way, way bottom, doesn't even make me understand the full implications of what it's saying. It's also very understated, and doesn't elbaborate how much of the semiconductor industry it's talking about, so I just see it as completely obsolete (even if it were updated), I just see it as a distant second to the current layout.
Now, in response to your points:
(1) Well, you answered yourself here, so I don't know what the question is. Regarding saying there's a precise description later about how the law works, I don't know why you point that out, or what you mean by it, because the purpose of HAVING an article is to give a precise description of something. The lead is different, because it summarizes (or just introduces) the subject, and the lead we've got does it beautifully. (now that it finally paints the right picture. Before, it seemed very off to me, hence my deciding to work on it)
(2) Q: Do we need to mention this before anything else? A: If you're asking me, the answer is a YES.
(3) Q: Is the fact that Moore's Law isn't a natural law worth mentioning? A: Your reasoning here is sounding like you think that everyone who drops into the article is going to *already know* what Moore's Law is, and not *need* to read anything. You said it's "so obvious in fact that I believe it's hardly worth mentioning", but this an encyclopedia article: you have to assume that the reader knows NOTHING, until it is stated. And the fact that it's that obvious to US Intel-knowing-people makes it darn worthy of being in the lead. So, to answer your question: Yes, I think it's EXTREMELY worth mentioning - and exactly in the way that it is - because it defines the very soul of the law.
(4) Q: Do we need to mention this first? A: I think this a repeat of 2.
(5) Q: Is Moore's law "just" a business practice? A: It is, and I can prove it: have the R&D departments of all chip-makers change their aim to developing much slower chips each year, and the law suddenly vanishes in its entirety. If the alteration of business practice can kill the law like that, then that's all it is. I think what Singularitarian means to be saying above is just that it's impressive to the point of being noteworthy that chipmakers manage to hit the mark each year. (remember also, though, that it's most certainly not ALL chipmakers who follow Moore's Law; some because they aren't ABLE to, and some because they aren't TRYING to. Moore's Law is quite limited in the number of companies that follow it... don't forget).
Anyway, no offense to anybody at all, if any can be detected in the above. (that's not what I feel in my 'expression' here) My bottom-line mentality is that I just have a very solid "comfort" with the way the lead is right now, for quite a number of reasons, listed above.
[ Dario D. ] 10:10, 17 September 2009 (UTC)

Dario, I appreciate your discussion but I disagree with you, and I agree with Gillingham much more. I find this to be worth digging into. I maintain that Moore's law is more than just a business practice. Moore's law predicts (rightly or wrongly) that it will be possible for humans to continue a certain business practice. That's different.

There are people (not necessarily me) who suspect that Moore's law is part of a larger trend towards order (Kurzweil's Law of Accelerating Returns) which may in fact turn out to be a law of nature. My point is not that this is a true statement, but that it is something people debate. Your statement that Moore's law is just a business practice is controversial and I agree with Gillingham that it shouldn't be in the first paragraph. Also, it shouldn't be phrased as a simple statement of fact. I think a statement like "some people argue that Moore's law is merely a business practice", together with a reference, would be better.--Singularitarian (talk) 01:35, 21 September 2009 (UTC)

I'm not sure what your logic is exactly here, because I find the premise of Moore's Law being a non-natural-law and business practice something that is deathly simple. If someone out there is thinking that it may be part of a larger trend towards order (sounds superstitious), it isn't currently anywhere in the article (at least that I can see), and even if it were, it wouldn't be carrying any real weight. It would certainly be nothing like the lead here that is defining the very premise of what this article is about. You mention that you don't agree, but I think you'll have to go over my other post, above, before I understand what you're standing on exactly. To me, the correctness (and relevance/weightiness/etc) of the current lead-section is far too simple a subject, so I can't figure out what the deal is with finding fault with it. lol, I swear these arguments are coming from the land of nit-picking... as if you guys are just really bored, and sit around trying to come up strange, plausible-but-not-reasonable POV's with which to see random things as wrong. (only talking about this current issue) - Btw, I did add a great citation not too long ago, reflecting both points in the line I added. (I thought I'd never need it, but oh well) I guess you didn't catch that yet.--[ Dario D. ] 11:52, 21 September 2009 (UTC)
I agree with Singularitarian that there is a content problem. The fact remains that Ray Kurzweil and others argue that it is more than a business practice. (Although even they don't go so far as to claim it is a natural law.) Because someone notable thinks so, we can't presume to have the last word on this. We are mere Wikipedians, and even if our logic is perfect and our conclusion is correct, we can't report that. We have to report what the literature says. If two sources say two different things, we have to report that. That's what WP:VER is all about. It doesn't matter what I think, or what you think. All that matters is what published sources say. The Singularity is Near is published and well known, so we have to give it a little consideration, even if we think that Kurzweil (as another editor said above) is a member of the "tin-foil hat crowd."
(Although, as I say, my own opinion doesn't matter, you can see a sketch of my own view here. I think your contrast between "natural law"/"business practice" is a false dichotomy and that you use "natural law" as a straw man, ignoring more considered theories of progress, such as Theodore Modis, as well as more popular theories of progress, such as Kurzweil. My point in separating the issues above was to make it clear that just because it is not a natural law, it does not follow that it is "just a business practice".) ---- CharlesGillingham (talk) 19:35, 21 September 2009 (UTC)

And another thing ...

After thinking about this a bit, I guess my main problem with the current lead is sort of "aesthetic" (in the words of User:Dario D. in the previous section).

The lead as it stands to today is simply bad writing. It doesn't flow logically or stay on topic. The second sentence is in the form of a rebuttal, since it uses the words "rather" and "however", but it doesn't provide any information about either the argument or its rebuttal; it just names them and quits. It discusses "natural law", but it doesn't say who considers it a natural law or why they might think so. It discusses business practices without explaining what those business practices are. The reader has to wait till the third paragraph to find out what "business practices" the second sentence was talking about, and natural law is never mentioned in the article again.

To illustrate what I mean, consider this, which has the same form but is about a different historical trend.

The Westward expansion describes a long-term trend in the history of the United States, in which settlers created new towns and cities in the western frontier of the nation. Rather than being a naturally-occurring "law" that cannot be controlled, however, the Westward expansion was effectively a form of population control.

Do you see what I mean? A reader who is not familiar with manifest destiny could immediately ask "who said it's a natural law?" The article hasn't brought up the issue yet, so it seems to funny to discuss with no preamble. Also, it doesn't explain exactly how it's population control. Consider this example, which stays on topic and flows coherently:

The Westward expansion describes a long-term trend in the history of the United States, in which settlers created new towns and cities in the western frontier of the nation. The trend began in the 18th century and lasted until 1890, the year which Frederick Jackson Turner termed the "closing of the frontier".

Journalists of the 19th century, such as John L. O'Sullivan argued that the Western Expanision was the consequence of a divine natural law, known as manifest destiny. However, historian Howard Zinn has argued that the westward expansion was a form of population control for the industrialized east. In order to avoid civil unrest, American lawmakers passed laws, such as the Homestead act, designed to move new immigrants and the impoverished away from overcrowded eastern urban centers.

This provides the next logical piece of info for a historical trend: when it happened. Another paragraph, a little further down, discusses the "cause" or "essence" (to use Dario D.'s word) of the trend.

In short, before we start talking about natural law, we need a sentence something like this to set up the issue: "This prediction has come true so beautifully, that nowadays we speak of Moore's Law as if it were a law of nature." (taken from Dario D.'s source). Otherwise, it's not clear why we're talking about natural law at all. Also, it makes more sense to talk about Moore's law being a business practice in a paragraph that explains what those business practices are. ---- CharlesGillingham (talk) 21:10, 21 September 2009 (UTC)

I also have issues with the opening paragraph. What does "effectively" mean? Either Moore's law is or is not a business practice, so let's just come out and say it and not hide behind this word "effectively". Gillingham, I like your example about manifest destiny. I'd like to see someone revise the opening paragraph.--Singularitarian (talk) 14:34, 19 October 2009 (UTC)
I'm a little busy today, but I will take another crack at the lead tomorrow. ---- CharlesGillingham (talk) 16:58, 19 October 2009 (UTC)

Power Consumption Section

The section on Power consumption is vague and outdated. I checked the source and it suffers from the same problems. I think he is trying to say power consumption of a microprocessor chip follows Moore's law, but he charts watt/cm2 for Intel chips. This does not really tell us anything and it looks very wrong, but I did not check since it is outdated anyway. Please delete or find a source that makes sense. —Preceding unsigned comment added by 69.125.146.118 (talk) 19:02, 31 January 2010 (UTC)

I agree, this seems wrong. It says node power consumption doubles every 18 months, correct me if I'm wrong but home PC's are nodes and they have not doubled in power consumption every 18 months. The source compares processor maximum power putting pentium 4's at 75 watts in 2001. If this was correct we'd have 5000 watt processors now... erm... don't think so. Muleattack (talk) 22:10, 13 May 2011 (UTC)

10 Nanometer Junctionless Transistor

This is huge! http://www.eetimes.com/223100050 If no one else has taken a shot at it, I'll try to update the "future" section over the weekend. —Preceding unsigned comment added by 75.71.50.97 (talk) 14:20, 25 February 2010 (UTC)


Continued trend

"The trend has continued for more than half a century and is not expected to stop until 2015 or later." The article says Moore introduced this concept in 1965. How can it had continued for more than half a century (+50 years) if as of today (2010) it has only been 45 years since the term was coined? Ptikobj (talk) 08:15, 17 March 2010 (UTC)

How could apples have fallen to earth for millenia though Newton's law was coined only in the 18th century? ;-) Thyl —Preceding unsigned comment added by 213.70.217.172 (talk) 09:19, 17 March 2010 (UTC)
The assertion in the reference that the trend starts in 1958 with the invention of the integrated circuit is specious at best, and no data is provided to support it. — Aluvus t/c 03:15, 24 March 2010 (UTC)
See the graph on the bottom of page 3 of Moore's original presentation of the idea. The data points in the graph appear to be approximately:
Year (data point) components per function doubling since 1958
1965 2^6.4 100 128
1964 2^5.65 50 64
1963 2^5 32 32
1962 2^3.58 12 16
1961 (no data) (8)
1960 (no data) (4)
1959 (no date) 7 or 12 (2)
1958 2^0 1 1 (Jack Kilby's first prototype had one transistor)
This is the data that Moore's law describes. It is very close to exponential, straight back to the beginning.
I hope this makes it more clear that the article is correct in claiming that the trend began in 1958, and that Moore's paper also makes the same claim. ---- CharlesGillingham (talk) 08:55, 24 March 2010 (UTC)
Add this as source: [4]
It even questions the beginning but:

The trend really goes back further. From the IBM 650 in 1956 to the IBM 370/168 in 1972 IBM mainframes went from 0.001 MIPS to 2.3 MIPS with the same doubling time of 1.5 years. Probably after the 68000 in 1979 micros had a lower cost/MIP than mainframes, as about that time mainframes did not keep up their doubling rate. So the "best price/mip trend" is really a 38 year trend of doubling every 1.5 years.


>System Date MIPs CPUs IBM-MIPs/CPU
>650 9/56 0.001
>7070 9/58 0.022
>1401 10/59 0.0074
>1410 10/62 0.0154
>360/50 4/64 0.178
>360/65 4/65 0.68
>360/85 1/68 2.4
>370/165 6/70 1.89
>370/168 8/72 2.3
>370/168-3 3/75 2.7
>3033 3/77 4.7
>3081-K 10/81 13.5 2 6.75
>3090/200 2/85 28.0 2 14.0
> >Datamation May 1985 page 90.
>Datamation Feb 1984 page 164.
—Preceding unsigned comment added by 69.232.205.64 (talk) 22:50, 1 March 2011 (UTC)


I added a sentence to the introduction that should help avoid this confusion in the future ---- CharlesGillingham (talk) 09:25, 24 March 2010 (UTC)
The data in the above table is historically incorrect. See http://ftp.jrc.es/EURdoc/JRC52422.pdf, p. 68, footnote 96. It says that Moore’s graph starts at 1959, with one component, and that, in 1959, two integrated chips existed: one by Kilby and one by Noyce. The former had two transistors and ten other components, and the latter had one transistor and six other components. ilkkaT 15:26, 13 May 2010 (UTC) —Preceding unsigned comment added by Ilkka tuomi (talk • contribs)
The table has no data for 1959, so how could it be incorrect about 1959? I stuck in your numbers. Jack Kilby's very first experimental attempt had one transistor, not that it matters. The point of the table is to show how Gordon Moore formulated the law by counting the doublings between 1958 and 1966. Some people were confused about the years to which Moore's law applies. --- CharlesGillingham (talk) 11:40, 14 May 2010 (UTC)
Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:Moore%27s_law/Archive_3&oldid=1048311025"