Slashdot Mirror
Intel Seeking Moore's Law Original Publication
ackthpt writes "Gordon Moore's famous prediction, labeled Moore's Law, was originally published in the April 19, 1965 issued of Electronics. Sometime since, he lent out his copy and it has never been returned. Intel would like an original copy of the now defunct magazine and is offering $10,000 for a copy, presumably in good condition. The story is carried on Reuters, and if you happen to have a copy (of your own, not stolen from a museum or library) you may contact Intel via eBay's WantItNow."
...it's unfortunately become a driving marketing factor for the industry...
...people realize they don't really need more than a 1Ghz to surf the web, send pictures, and listen to music...
...Moore's Law was coming to an end based on simple technical limitations...
Huh? These three points don't mesh at all. A driving marketing factor -> people don't need that much power -> but it doesn't matter because it's coming to an end.
If you can't imagine the use for more processing power, then you're not very imaginative.
Processing power is a remarkable thing - you're talking about 1Ghz as being a pedestrian, adequate level of computing, yet you in a prior life (or rather prior year), back in 2001, were undoubtedly saying "Oh who'd need these crazy 1Ghz processors? A 300Mhz is all anyone would ever need...". Even the luddites somehow pull their requirements forward to be just behind the curve, and I've been hearing the same "who needs more than X" mantra quite seriously since the 386 days. Some people never learn from history though.
Ooh, I'd love to have a 32GHz machine... it would make my acoustic modelling here at work that much faster... I occasionally get run times measured in hours on a 3.2GHz P4, and when it's an iterative process, requiring manual tweaks between runs, it gets kinda tedious, not to mention time consuming! And I imagine there are plenty of other scientists / engineers / economists / mathematicians out there who'd appreciate a ten-fold increase in speed, for an affordable price.
Oh, yeah, and maybe then I'd be able to run Operation: Flashpoint (http://www.flashpoint1985.com/) at maximum detail settings, too! It brings my Athlon 3000XP & 9800pro to its knees when I crank up the detail (especially if you push view distance beyond about 2.5km)
They might be marketing driven, but on the other hand the processing power nowadays really extends computers. Use hardware from '95 and you'll soon realize that you can't listen to mp3s in real time or that you can't encode some video within a week. I guess that we all agree that every modern computer mostly wastes its cycles, but sometimes it's rather handy to have that extra power at your fingertips.
If they need some stupid "law" to follow it's allright to me.
There is just one thing that bugs me since years: That every new gerneration of chips consumes more power in order to fulfill Moore's prophecy. But I guess we can only blame the consumers in this case.
I don't read replies by ACs.
I guess I'm just finding it difficult to imagine what I would ever need, say, 32Ghz for, other than gaming--which would be what my ultra-hip game console would be for. Thing is, computers are still advancing at a rate that console have a problem catching up with, due to the fact that every aspect of the hardware has to be released at once and, of course, that they can't come in pieces developed seperately and be interoperable between wildly different hardware (all for obvious reasons due to the nature of consoles; I'm really just imperfectly repeating the obvious).
So, for truely impressive games media-wise (alas, the more complex they are technically the less time is able to be spent directly on gameplay and etc, but that's another issue) computers have for quite some time been far better, and probably will remain so for quite some time hence. Consoles sometimes nowadays seem nearly comparable, at least in some cases, but then you look at the abysmal resolutions and you realize that they're cheating performance-wise.
Other than that, I enjoy my rather fast computer (far beyond any console currently available) for many things, for example quick encoding, compression, and reformatting of video and audio (usuallly for entirely legitimate purposes, oddly enough!). As media types are advancing, it's nice to have the hardware to keep up. And to be able to digitally record, edit, and redistribute a movie "filmed" (a misnomer now) by myself, at quality certainly below DVD but superior to VHS and anything concievable at a lower level of technology . . . no, I'm glad that computers are advancing at the pace that they are, and I certainly do find the use.
That being said . . . about two years ago, I became unable to really use the computer that my family was using, a state of the art machine but it was constantly under use by my other family members. Being still in high school at the time, I certainly needed a computer. Especially for chatting and browsing late at night, and at those times I wouldn't be able to make my way through my unbelievably creaky house all the way to the downstairs anyways. So I dug out an old computer of mine, a Pentium-S 100-mHz if I recall correctly, with a massive 8MB of ram or something equally woeful.
And you know what? Armed with programs made in the era it was from (which was a bit tricky to find old ones that would interoperate with the state of things nowadays, but it's possible), it performed quite adequately. I even abused it with programs that should have been far beyond it's ken, but it still trotted along fine (until one day I accidentally destroyed the HDD, but that's another story). And so this computer from the mid-ninetys was easily good enough and functional in the modern age.
In other words . . . I disagree with parent. And then, on further reflection, I completely agree. I hold two conflicting opinions.
I remember sigs. Oh, a simpler time!
I agree with you to a point, but in 2001, I was never saying 300mhz was all anyone would need. Applications like Office were still slow and memory-hungry. Booting up Windows was a minute-long affair.
That has changed. 512MB of RAM and 1Ghz are a very common baseline now. Office runs just fine on it. E-mail clients run just fine. Web browsing has never really been a system-hungry activity. Gamers are a special niche; most computer users have no need now for more than 512MB of RAM and 1Ghz.
Honestly, can you tell me what they'd need more for if all they're going to do is type documents, send pictures, and surf the web? A friend of mine is still running his 800mhz Powerbook at work with 128MB of RAM with no complaints.
Look at it this way. Cars used to be really slow, and you had to hand-crank them. They got faster and faster and more practical with each decade. Now you could build a car that could go 500mph, but nobody needs it except race car drivers. The only thing that's really changed with cars is more efficient fuel consumption and various niceties like stereo systems and computer navigation. And yet, I'm still getting along with my car made in 1993. Before that, a car made in 1984. A friend of mine still drives a 1960s pickup truck. He simply doesn't need more than that.
I think computers have hit that plateau. I also think that's part of the motivation behind rearchitecturing all of Windows Longhorn so that it's all "managed" and requires an extra layer and more hardware resources to run it all, and therefore a new computer purchase much to Intel's delight. In the Apple world, Mac users hang on to their machines longer because they're not living in a Moore's Law-inspired annual PC upgrade cycle that takes your money every year.
Just my $0.02.
Moore's Law has "nothing" to do with clockspeed?
Moore's law describes component integration on integrated circuits that are economical to manufacture. This results in, among other things, increased processor speed. Generally speaking, Moore's Law has been adopted as an observation on general computing power.
This is like the people who desperately argue that "hacker" originally meant something else, and that we should all use "cracker" instead. You and I know what we're all referring to, so the argument is just a nuisance.
I think there will always be a big demand for faster processors, but I don't think there is a need for them. I also don't think that the main focus should be on how to make faster things fit in smaller spaces. They seriously need to do something about this heat problem. 70C for a 775 Prescott chip is an O.K. temperature. That's a bad thing. Let's put some money into fixing the problems with the technology.
To me, making things faster without improving any other aspect of them is like making a car that goes 300km/h but has no additional safety features.
Has clockspeed doubled in the last 18 months? No, not even close.
Will it continue to ramp up as it has in the past. Probably not.
Has the number of transistors doubled every 18 months? Yes and it will continue to do so for awhile yet. Moore's Law is valid and will be valid even if there is no clockspeed increase, until we stop doubling transistor counts. You make the mistake in directly tying increases in performance to increase in clockspeed, which is a an oversimplification of what goes on in a cpu.
As to what the average person calls something I could care less, Moore's Law has always pertained to transistor counts.
Performance isn't tied down to clockspeed.
"In physics, do we say that force is about equal to mass times acceleration?"
No, but you should.
Have you personally observed the forces involved in the acceleration of a black hole from 0 to c in less than 10^-23 seconds? No? Then how do you know your "law" holds?
There are several photocopies and digital copies that are floating around (even someoen ebaying a PDF version for 50 bucks, and he has bidders..??) So all evidence won't be completly destroyed.
I am very certain that if all I had access to was 1Ghz with 512Mb RAM that I would find some way to make do... And yes I am a gamer, so I'm not gonna go into that side of requirements... but I have a gaming system that I use everyday for routine activities as well, 3Ghz, 2Gb ram, and I can tell you I very much notice the difference.
;) )
As humans we are creative enough to find more ways to use resources than we have resources to use... Thats why the next processor/mem/speed combination is never enough for very long. We find ways to expand until we are constrained by our resources.
my personal computer is so much better than these (2-3ghz?--system info applet is disabled)p4 512 ram that I actually almost always remote desktop into my own machine, and take the network latency hit because my current computer usage style swamps the systems my university provides.
Right now I have 11 explorer windows open, 2 e-mail, 4 messenger windows with one tabbed with 10 internal windows 4 firefox windows with no less than 8 pages open in each, 1 dos prompt, 2 ssh windows / filetransfer windows 5 notepadwindows 3 pdfs 2 visual studio windows
All of this just for my work (well minus some of the messenger windows). Then I also have a few random apps like skype, bittorrent and itunes, copernic desktop search as mentioned on slashdot, daemon tools and mcafee
Now minus the random crap at the end... all the rest is continuously active... + is my friend 1600x1200 is my pal -- sometime soon I'll be getting another monitor...
The point is that though office, e-mail and webbrowsing a single page all works perfectly fine on the baseline system, people will only be satisfied with the baseline for so long... people's usage habits change as the become accustomed to a certain technology... I wouldn't have started using this many programs if I had to still do it on a 400mhz as we moved up in systems I kept modifying my usage patterns such that after the initial newness of a system wears off I find myself at the limit of its capabilities...
I disagree with your thought that computers have hit that plateau. The car analogy is flawed in this case... I think that along with a gradual modification of the populace's general usage habits, more and more applications will be made available that need more resources. --This is happening already. With more resources available, people will think of more brilliant uses for those resources... AI problems and other complex problems will find more and more solutions in the available hardware... but that hardware will have to grow to fully take advantage of it...
(my $0.02 for yours
Gravity Sucks
They're not the cutting edge. Hell, I'm not, but I see a need for speed because of: mp3's, video-on-demand, podcasting, voice-recognition, rippin' dvd's, capturing TV (myth, xptv, or whatever), centralized media and multiple remote players, kids doing homemade animation, gaming, backups, making backups of dvds so the 2-year-old doesn't destroy the original, advanced video processing, sound editing, home photography archives...
Nah, I don't need a gigabit net, firewire, raid or fast computers. This here 1-mhz Altair, wordstar, and 8" floppies will do me just fine. Together with a daisy-wheel, I can do all the writing I want. Really.
Still to come: videophones, real-time avatars, bespoke animation/video, more on-demand video/audio (including education and games), always-on videoconference ability, trivial offsite backups/redundancy, depth-of-field or other non-video data added to video feeds, any-to-any video feeds (think n-megapixel streaming cameraphone), realtime data analysis on problems that currently are out of reach, even broader upheavals between mainstream-media and blogs/indy musicians, etc.
Every time you give me more power, I'll find problems worth solving and places to use it. I used to slip a digit in some finite element work and take puzzles from 40-hrs of cpu time to unsolvable. Given a few more years, my old work will be running at 30 frames a second.
Right now, a pic of the Power5 chip is pinned to my wall: 8 cpu cores, 4 mmus, 144mb cache, one chip/die. Ads say this baby scales up to 16-ways for 128 cpus possible, at 2ghz. I say it's just a good start...
Look at it this way. Cars used to be really slow, and you had to hand-crank them. They got faster and faster and more practical with each decade. Now you could build a car that could go 500mph, but nobody needs it except race car drivers. The only thing that's really changed with cars is more efficient fuel consumption and various niceties like stereo systems and computer navigation. And yet, I'm still getting along with my car made in 1993. Before that, a car made in 1984. A friend of mine still drives a 1960s pickup truck. He simply doesn't need more than that.
I think computers have hit that plateau.
Your analogy is flawed. Computer speeds have to keep up with the increasing requirements of software. As software gets larger and does more, processors need to become faster in order to present the same speed of interface to the user. Cars, on the other hand, don't have changing requirements. The roads they run on today are the same roads as fifty years ago. The only reason we'd *need* faster and more powerful cars would be if the roads kept getting steeper and steeper. Since that's not the case, cars don't change much and you can get by with an older car because it runs on the same road as a modern one.
So if you're happy running the software you did a decade ago then you don't need a faster computer. But if you want your software to do more and more, you're going to need a faster processor unless you enjoy waiting around longer and longer.
Want to improve your Karma? Instead of "Post Anonymously", try the "Post Humously" option.
This is a concept I am going to try and explain in simple and eloquent terms that even the most humorless slashdotter can understand. I know, being a crowd of logically minded folk the responses tend to be well thought out and analyzed from every angle to ensure that nothing but the highest quality ideas and posts make it to this site. With that in mind, I do believe the original poster was attempting this odd thing known as 'humor' This is an act where one makes a statement that is very outlandish and silly. It is typically an attempt to make others 'laugh' or 'smile'.
Well then. I feel better now.
Jeremy
They have a copy of the paper, but they're willing to pay 10 grand for an original hard copy. Why? Do they think that the surviving digital copies have been altered? My guess is that it's some kind of weird publicity stunt. In a week they'll say, "we gave John Doe $10,000 for helping us prove that our processors beat Moore's Law." Reminds me of nVidia bragging about the same thing when they released one of their GeForce GPUs. BTW, if anyone can find a copy of the original press release, I'll gladly pay them $10,000. ;)
Yeah, but that doesn't take into account the brick wall that is "IP" law, which is just now starting to prevent most of the previous works from being built upon.
I have to disagree a bit. Back when the PII 300 was king, there was still a lot of 'waiting' going on. I'm sure a lot of it may have had to do with memory limits back then (remember, Win95 can't address more than 64MB efficiently), but overall everything was still somewhat slow. We all felt that things could go faster.
;-)
Also, keep in mind that when the high end was 300MHZ, the low end was still in the 486 range. That was one hell of a defecit between the low and the high. Low end machines could barely run the latest office apps.
Fast forward to now. The low end box out there is roughly a 1GHZ machine with 128-256MB RAM. Let's say 256MB is the low end for argument's sake. The high end box (in a corporate or home environment; we're not talking video rendering here) is in the 3GHZ range with 1GB RAM.
Now, take your average home user or administrative assistant, and have them speed test each machine for a day. Then ask them which was faster overall. I'd bet that many of them can't tell the difference. Even those that could would most likely tell you that the difference was negligible, once all the apps load into RAM.
My point is, the low-to-high end performance gap now compared to 2001 is exponentially smaller. We really are reaching the point where a two or three year old computer is more than enough for the majority of computer users out there.
I think if you could quantify the 'usefulness' of the personal computer, contrast it with the machine's power on paper, and put it on a reverse timeline, you'd see something on par with Moore's Law, only in reverse, and much more accelerated.
Hell, I think I'll chart it all and release a paper on the subject. I'll call it Jerky's Law, and will someday offer a reward for the original slashdot posting
I recall seeing amazing programs running in 16k of RAM on a 2Mhz Z80. What happened to the brilliant software designers of that era? They're sure not working on today's platforms.
Software is designed to do a lot more these days. Team sizes have gone up significantly. Unfortunately, when you have 10 or 100 (or 1000) people working on a piece of software, it can't be made as "tight" as one person trying to squeeze it all onto a machine with 16k of ram. The interface artist wants to include transition animations to give the users a spatial mapping thereby increasing the ease of picking up an application. The Logic coders want the applications to serve the user only with the data that is relevant to them. The backend coders want the application to run on four platforms, supporting any number of hardware and software interfaces. And management wants all of this to be done cheaply and quickly. Alone, any of these things could be optimized well, but integrated into a massive platform which outsizes one person it gets difficult to find and fix all of the bottlenecks. Add to that the additional layers of abstraction that have made coding much, much easier than it once was, and you have a different beast.
Plus, those 16k of ram applications had terrible interfaces. I remember my father worked on a text editor which shipped with a manual the size of a brick. It had no help functions, no obvious mapping between keys and input, if you gave it input it didn't understand it just didn't do anything... It was like VI, but less user friendly.
If software ingenuity progressed anywhere near the rate of hardware, we would have infallible voice and character recognition, true A.I. and the concept of computer crashes and security problems would be a thing of the past.
We have great voice and character recognition (Dragon rocks), but without a lifetime of experiences powering it they fall short of what people expect. "True" A.I. is impossible until we understand how Natural Intelligence functions, a hurdle neuroscientists have yet to overcome. Crashes and bugs are a factor of any engineering project, be it computer programming or building bridges. And as computer power has increased, so has the power afforded to hackers. But it is a lot harder to get a good virus going these days.
But what do we have? Abstraction layers that let you program a 3D game or a 2D word processor in an afternoon. Inline spelling and grammar correction. Graphical processing suites that can instantly change a photograph into a convincing watercolor. 3D CAD applications that can create functional hardware for extreme circumstances, and rapid prototyping machines to build parts. Atonomous agents which look for RSS feeds of desirable files, and distributed download clients which gather them. Unlimited levels of Undo. 100% fault-tolerant remote communication between any two machines in the world. Machines running virtually inside of other machines. Bayesian spam filtering. Heck, if we're talking 1982 computing, add the concepts of protected memory space, (gasp!) multitasking, and multithreading to that list. And audio / video compression and streaming, or for that matter graphics and pretty much everything that has become standard for the internet. Searches we perform routinely today would have been considered intense data mining operations 20 years ago.
While it may not have been the direction our forefathers believed (promised?) computing would go, it has still been a very interesting past 20 years.
The ______ Agenda
I recall seeing amazing programs running in 16k of RAM on a 2Mhz Z80. What happened to the brilliant software designers of that era? They're sure not working on today's platforms. ...embedded platforms, and the system/toolkit libraries we all love and use. Some of that stuff is still assembly optimized. As for RAM, in games graphics, music and game state take far more than any "code" does. In business apps, the data sets do. Nobody cares if you're using 16kB or 1600KB if you're working on 10000kB of data.
As for CPU speed, most people have far more than they need. But compare the cost of a CPU to developer time - it simply doesn't work out. It is much much cheaper to allow developers to use large abstractions and system libraries (which eventually get very clean and bugfree), than it is to use slower CPUs. Again, except in embedded markets, and even that's changing.
Kjella
Live today, because you never know what tomorrow brings