Sign in with
Sign up | Sign in
Your question

The Future of the CPU?

Last response: in CPUs
Share
January 17, 2008 5:04:41 PM

Hi guys, I have been briefly trying to research the future of CPU's and personal computing beyond die shrinks and I am not finding much so I wanted to stop in here and ask if any of you could point me in the right direction. We have 45nm cpu's at the moment and the laws of physics will not allow much beyond about 20nm so it becomes obvious that this cycle of die shrink fueled performance increases will be comming to an end probably around 2015-2020. What lies beyond? What current research is underway? Are Intel and AMD going to change from pioneering growth companies to benign manufacturing plants?

I have read about quantum computing but I don't yet see how that applies to personal computing. I appreciate any direction you can give me.

Thanks! :hello: 

More about : future cpu

January 17, 2008 5:38:16 PM

I'm not really sure, but I think I've heard a little bit about fiberoptics, or using lasers or soemthing. I don't really remember.
January 17, 2008 5:43:36 PM

Yadge has read or heard the same thing i have read about. It was a few weeks ago, so i dont remember the link. It was about a large company, maybe IBM, working with lasers. Using them similar to fiber optics. They were getting some crazy performance, like a 250x faster. Sorry for the terrible paraphrasing, but it was before the holidays.
Related resources
January 17, 2008 5:49:06 PM

50bmg said:
Yadge has read or heard the same thing i have read about. It was a few weeks ago, so i dont remember the link. It was about a large company, maybe IBM, working with lasers. Using them similar to fiber optics. They were getting some crazy performance, like a 250x faster. Sorry for the terrible paraphasing, but it was before the holidays.


I don't have the link..

But yes, it's a move away from electrical and to fiber optics inside the computer. It is much faster and would generate less heat. Electricity does not move at the speed of light.

Should be interesting in the next few years where we see things headed.
January 17, 2008 6:04:11 PM

Also check out leech brains.

You think Im kidding, but some scientists have used leech brain cells to process information.

If you are trying to get info for a report or something, it could provide some engaging content.

Also look into liquid polymer memory.
January 17, 2008 6:19:33 PM

rallyimprezive said:
Also check out leech brains.

You think Im kidding, but some scientists have used leech brain cells to process information.

If you are trying to get info for a report or something, it could provide some engaging content.

Also look into liquid polymer memory.


^Living neural networks, quite a bit of research is going on with neurons on biochips. Just not quite sure if cells can ever replace silicon/solid state due to their length scales - the required living chips would be much larger than the processors we know today. Neurons are on the order of 10s of microns whereas their processes (axons/neurites) can be sub micron in width and 100s of microns long (even cm).
January 17, 2008 6:47:40 PM

Ballistic Deflection Transistor (BDT) processors would offer terahertz speeds at a tenth the heat generation and power usage. Unfortunately they have a bunch of idiots trying to turn it into a completely new architecture that will favor investor companies. Since it is going to be using certain technologies in partnerships with certain companies it will not only be getting snuffed out by Intel and AMD but now just about every chipset maker out there.
January 17, 2008 7:59:15 PM

I watched a ultra nerdy nerdy show the other day regarding absolute zero (-273K)

The interesting thing is at -272.999999999K a matter is formed (Bose-Einstein condensed ) they showed Bose-Einstein condensed can slow down light without manipulating it, also they briefly mentioned the possibility of its use for processor technologies.

**to make a long story short Bose-Einstein condensed is Atoms moving very slowly as one unit apposed to individuals…. Atoms lose their identity.

Of course the use of Bose-Einstein condensed in the future is mostly speculation at this point, it surely will find its way into our processors at some point….When, who knows?

If you google Bose-Einstein condensed; processor you can find a lot of interesting things about it.
January 17, 2008 8:11:24 PM

grieve said:
I watched a ultra nerdy nerdy show the other day regarding absolute zero (-273K)

The interesting thing is at -272.999999999K a matter is formed (Bose-Einstein condensed ) they showed Bose-Einstein condensed can slow down light without manipulating it, also they briefly mentioned the possibility of its use for processor technologies.

**to make a long story short Bose-Einstein condensed is Atoms moving very slowly as one unit apposed to individuals…. Atoms lose their identity.

Of course the use of Bose-Einstein condensed in the future is mostly speculation at this point, it surely will find its way into our processors at some point….When, who knows?

If you google Bose-Einstein condensed; processor you can find a lot of interesting things about it.


hm...wouldn't that be C not K :p 
it would be about 0.0000000000000001K, but anyway, interesting
January 17, 2008 8:11:47 PM

Aweseom guys! You have given me plenty of key words to keep me busy for awhile. I have hit a lull at work here and I have some time on my hands so I honestly just wanted to research for entertainment. I am an electrical engineer so this type of stuff remains interesting to me. My father also holds stock in a CPU company so I thought the research might provide direction there too.

Thanks again!
January 17, 2008 8:29:37 PM

monst0r you are absolutely right... I was too lazy to look up the 0.0000000000000001K so I took a guess in Celsius... I suppose I should have put -272.999999999C(approximately)
January 17, 2008 8:49:30 PM

cnumartyr said:
I don't have the link..

Electricity does not move at the speed of light.




Strictly speaking nor does light, when travelling down a fibre optic cable :) 
January 17, 2008 9:06:18 PM

I watched the same nerdy show and had the same thought as Benzene...

Slowing light down is no big deal. Light always travels slower when traveling through a medium (read: not a vacuum)

At least thats what I remember from Physics class
January 17, 2008 9:53:36 PM

50+ years time: Organic brain computer! Eventually leading to the use of networked human brains to create large self maintaining networks that will continue to expand and eventually remove individual identies into one global collective conscience that will continue to advance at an expinential rate...

I for one welcome our new biological super computer overlords.
January 17, 2008 10:29:18 PM

I was thinking quantum computers.

You know, instead of the usual 0 and 1 binary digits, quantum pc uses electron spin, which is either up or down.
January 17, 2008 10:35:46 PM

while they may not be as far reaching as you want, there are acouple of technologies that are coming to the front over the next 10 years to keep silicon based transistor technology going.

tri gate transistors
carbon nanotubes

Theres numerous other new great ideas being explored at the graduate school level and thats where all of this ends up coming from.

good luck

January 18, 2008 1:30:18 AM

time to start investing??
January 18, 2008 2:07:47 AM

how about you guys just worry about now, and not about the future. yes there is a lot of research on all of the possibilities of future technologies. We have no idea whats going to happen in 10 years. yes there are always stuff in the news, tv, movies, magazines, but what about the research that you haven't heard about. I don't know if you guys have realized that this entire world revolves around some sort of currency. Just cause it is the best product doesn't mean it will make. the person with the most money wins
January 18, 2008 2:49:39 AM

crazywheels said:
how about you guys just worry about now, and not about the future. yes there is a lot of research on all of the possibilities of future technologies. We have no idea whats going to happen in 10 years. yes there are always stuff in the news, tv, movies, magazines, but what about the research that you haven't heard about. I don't know if you guys have realized that this entire world revolves around some sort of currency. Just cause it is the best product doesn't mean it will make. the person with the most money wins


Dude, you completely missed the purpose of this thread.

Go discuss money driven objectives in the Intel vs AMD threads.
April 13, 2009 11:32:49 PM

I'm gonna bump this thread. So after a year do we have any progress? Does anyone know the ETA for one of those claimed "revolutionary" CPUs to appear on market?
a b à CPUs
April 14, 2009 12:25:01 AM

I heard about using a new material, graphene, derived from graphite instead of silicon, which could theoretically rach the Tera Hertz speed (1000 GHz!!!). I think Intel is into it... as well as IBM.

Another thing I read about is probabilistic processing, in which answers are "guessed" instead of processed completely, I think. These CPUs should consume much less and work much faster... but I don't quite like the idea of having a piece of hardware that may guess incorrectly...

http://technomindz.blogspot.com/2009/03/graphene-proces...
http://www.electronista.com/articles/09/02/08/rice.univ...

Lol, I've calculated that, according to moore's law, by 2030 CPUs will have KILO-CORES!!! And by 2050, they'll have MEGA-CORES...
a b à CPUs
April 14, 2009 1:20:48 AM

grieve said:
I watched a ultra nerdy nerdy show the other day regarding absolute zero (-273K)

The interesting thing is at -272.999999999K a matter is formed (Bose-Einstein condensed ) they showed Bose-Einstein condensed can slow down light without manipulating it, also they briefly mentioned the possibility of its use for processor technologies.

**to make a long story short Bose-Einstein condensed is Atoms moving very slowly as one unit apposed to individuals…. Atoms lose their identity.

Of course the use of Bose-Einstein condensed in the future is mostly speculation at this point, it surely will find its way into our processors at some point….When, who knows?

If you google Bose-Einstein condensed; processor you can find a lot of interesting things about it.


Riiight point K at your doorstep? Never gonna work.
a b à CPUs
April 14, 2009 10:30:16 PM

amdfangirl said:
Riiight point K at your doorstep? Never gonna work.


But you'd be close, if you lived out in the Oort cloud :) . Maybe 5 degrees K - a little refrigeration and you're there!

The ballistic transistors article looked interesting - sorta like the integrated vacuum devices of the 1980s which resembled vacuum tubes operating at 5 volts instead of hundreds of volts, and used electron emitters that were basically pyramids ending in a point a few mils (millionths of an inch) across so as to generate the necessary voltage gradients to operate at such low voltages. However the vacuum devices used a grid-like structure to control the electron flow to a fixed collector, not steer it to different targets. Mainly of interest to the military and space applications since these devices were inherently much more radiation-hardened than semiconductors.
April 16, 2009 12:50:10 PM

Theoretically optical and/or biomachinery is the future, but part of me wonders if that stuff will ever even be necessary outside of the scientific community. Aside from hard drives, specs and performance increase seem to be slowing down and, perhaps more importantly, the rate at which minimum specs for games and stuff increase also seems to be slowing down. Could we be approaching the point where the limiting factor is software development and we go from a performance arms race to focusing on features and selling 'old' chips?

Perhaps computing hardware will go the way of cars and appliances where simply having one is what matters, rather than how new it is.
a b à CPUs
April 16, 2009 12:51:25 PM

Well we'll see, anything new will appear in server first.
April 16, 2009 1:37:27 PM

Yes. Fiberoptics. Now in a CPU, copper is used to transfer data. This greatly limits the no. of transistors that can be built in. Intel is already working on a CPU that sends data through light. This is not only faster, but also decreases the size of the CPU and will result in greater performance.
April 16, 2009 2:07:15 PM

Is nanotechnology going to assist CPU technology in the future? Or is fiber optics a more practical development right now?
a b à CPUs
April 16, 2009 3:01:45 PM

Plenty of Science labs with quantum computers etc. trying out these ideas.
April 16, 2009 4:47:08 PM

Fiber optics plays a major role. It was like Intel kept on increasing their clocks until heat became an issue. But on the size issue, I just found the article I had read a long time ago! Yay! Atomic scale computing! Here you go. http://www.foresight.org/nanodot/?p=2926

On a side note: AMD fangirl, WOW!!!! That's a first. 14 yrs!! And an eternal poster. Miracles never cease.
a b à CPUs
April 17, 2009 9:16:22 PM

rags_20 said:
Fiber optics plays a major role. It was like Intel kept on increasing their clocks until heat became an issue. But on the size issue, I just found the article I had read a long time ago! Yay! Atomic scale computing! Here you go. http://www.foresight.org/nanodot/?p=2926


Rags, that was a very interesting article - thanks for the link :) . Reminds me of a patent I once saw, using similar STM techniques, of moving atoms around on a nickel substrate to form logic circuitry. The atoms "stick" in place due to the Van der Waals forces, except certain atoms exhibit much less stickiness. So to form a flip-flop, you make a loop of sticky atoms, but leaving a gap at the end of the loop, and using a loose atom in the loop nearest the end. You then send an electron merrily buzzing along the loop, and when it gets to the end, its EMF causes the loose atom to move out of the loop, so that when the electron bounces off the gap to buzz out of the loop, it gets stuck by the new gap caused by the atom moving away..

April 18, 2009 12:25:20 AM

^Hmm.. Intel can research, sure, but I can't see such CPUs in the market for years.
a b à CPUs
April 18, 2009 12:26:33 AM

Can I say, my Science skills are no way good enough to understand the concepts behind it.
a b à CPUs
April 18, 2009 2:41:56 AM

amdfangirl said:
Can I say, my Science skills are no way good enough to understand the concepts behind it.


Most science is not that hard to understand, except maybe for the 'spooky" quantum stuff. If science or engineering is your interest, study hard and get into a good college, & you'll pick it up. And maybe start reading some science magazines, such as Scientific American, or the NZ equivalent, in the meantime :) .

Also, using "Oort cloud" in casual conversation is good :D .
April 18, 2009 6:51:34 AM

I still can't believe 14 year old girls post in this forum.
And I agree with fazers. Science is not hard as long as you don't read it in school. I don't know about you, but I never used to like to study Science in school. Preferred to do it by myself.
!