October 26th, 2004, 10:27 PM
With Intel announcing they are not going to try to manufacturer a 4 Ghz chip and AMD not planing on doing so either one has to ask, have we reached the GHz limit? The push is now towards milti core chips and 64 bit. I think this is long overdue, the GHz race has just produced more heat and less stability.
I think GHZ of conventional proccessors is approaching the limit. I mean, there are only so many wires you can cram in a 1in x 1in square. However, I am also hoping that there is a breakthrough in quantum computers.
Instead of running on traditional binary, (on and off), a quantum processor uses all the quantum states (I think it is like 2^500, but that seems wrong to me). Obviously information can be processed incredibly faster uisng this methon. Kilobytes (possibly MB, I didn't do the math) could be reduced to a single byte using this method.
This would revolutionize computing, and would make many, many things possible. Imagine how fast any encryption algorithm could be cracked. This would seriously mess with the world in ways I couldn't fathom.
Google for more info on quantum computing.
Very close to an analog computer or optical analog computer, much like the human brain. Sounds like the next evolutionary step.Originally Posted by sploit
We have certainly hit a heat barrier with the current technology. But heat can always be dealt with insome way. But I read Moores law only can be fulfilled for about 600 years before it impossible to go any further with ANY technology.
http://arxiv.org/abs/astro-ph/0404510
It will be very difficult to put more transistors on a chip in ten years or so. By that time the transistors will have become so small random, quantum physics based effects, start to dominate and the transistor no longer will work. Ofcourse you can opt to go for a bigger chip but that isn't without it's problems either. One of them would be getting the signal from one place on the chip to another place on the chip in time.
On the chip frequency. Yes I think we'll see chips grow less in clock speed than before. If you looked at the Alpha roadmap as Digital saw it, you didn't see much increase in clockspeed. The design focus was for the most part on doing more work in parallel. The Alpha EV8 would have implemented highly effective SMT, but alas Alpha was killed by Compaq just months before the merger with HP was announced
We could of course always figure out a way to stack the chips directly on top of one another, making in a sense a cube.It will be very difficult to put more transistors on a chip in ten years or so. By that time the transistors will have become so small random, quantum physics based effects, start to dominate and the transistor no longer will work. Ofcourse you can opt to go for a bigger chip but that isn't without it's problems either. One of them would be getting the signal from one place on the chip to another place on the chip in time.
On the chip frequency. Yes I think we'll see chips grow less in clock speed than before. If you looked at the Alpha roadmap as Digital saw it, you didn't see much increase in clockspeed. The design focus was for the most part on doing more work in parallel. The Alpha EV8 would have implemented highly effective SMT, but alas Alpha was killed by Compaq just months before the merger with HP was announced
Heat is still a factor.
They need a way to capture the heat escaping and reuse it to power the chip. Then it would be self sufficient, almost
So that's what the Borg got started
LOL,So that's what the Borg got started
It makes sense if you think about it, having a three dimensional silicon cube would allow for a far greater transistor count and would solve most of the problems of distance inside the chip. Heat, now that's another thing.
Wasn't it IBM that made an optical memory cube?
Someone should invent a new kind of heat exchanger that should me more eficient. That should solve the problem.
New materials are also coming : SiC is going to replace the common SiO2
This together should let us have faster PC's
Kevin
what if we were to use a refridgerator unit, attached directy to a heatsink, it would be much colder than a standard fan, but still nothing compared to liquid nitrogen cooling
The heat pipes really work well. One of the links at the bottom of the page is a company that makes the LED lights and also they do cooling.
http://www.enertron-inc.com/
The pipes pull the heat away from the base using a liquid filled round pipe (oil?). The end result is far better heat transfer then just a large heat sink. Pretty much everyone does this now especially with the blade servers.
An air conditioning system is very complicated and requires to much space. Maybe a liquid refrigerating system could be useful, but whe don't know much
Water cooling helps some. It can drop your cpu block temp by 10 degrees C or so. While that is not a huge drop it does show that those now with temp problems can find a way around.
One of my computers at present is water cooled and has been in operation for several years without leak issues. About the only thing needed is cleaning the radiator fins, cpu cooling block, and refilling the water tank that loses water level because of evaporation. I am happy with the results of the cooling and it does enough for what I need.
Kind of cool, so long as it doesn't leak. Water is conductive, liquid metal is another story.
I think now it time to make more efficent chips. It is great to have all this clock speed but a AMD and Apple have been trying to say for years it is how you use it. I think multicore and 64 bit is now going to start taking off. I think companies are going to move past clock speed and start advertising effieceny. I think how much heat it make and how much power it uses are going to start becoming selling points.
I read an article recently in Scientific American about computers based on grids of nano-wrires, which could be considerably faster than modern transistor-based chips. I'm pretty sure it was theoretical, and that issue has now gone AWOL, so I am unable to give any more details. Sorry.
I've read some really sick things about quantum computing - look it up if you haven't heard of it, you'll be amazed at the projected speeds. The projected leap in speed for a quantum computer compared to the fastest computers that exist today is unreal.
I tried to write an assembly line to pipeline analogy, but there's just too much to go into without writing a full essay on this. The short and sweet is that cycles per second aren't everything, there are other ways of improving speed. The latest, greatest, processor from Intel, the Core 2 Duo, is only about 2.1GHz, but it's much more powerful than its 3.2GHz predecessors.
The other thing someone posted in this thread which I have to correct; quantum computers will not be used for general computing! They can simulate quantum physics very quickly, and factor primes, but they can't be used for playing Quake 4 or running a spreadsheet. (At least, not nearly as well as our silicon based computers.)
I think high performance software will become more parallel whereever it can, and chips will use less and less power and get more and more portable; and most importantly I think we will see interpreted/very high level languages completely replace low level C type languages for all but the most processor intensive tasks, but I think we'll start to see the upper bounds of performance in a few years.
Is this so terrible though? I've got a 1.5GHz Celeron M and I simply don't need anything faster. I know you can say "well back in the days of DOS we didn't need anything faster", but I can't see anything revolutionizing desktop computing any more. I hope I'm wrong, but I don't know where it's supposed to go from here.
im running a 3000+ amd 64 in my pc, but my motherboard can take dual core processor. if i run the fastest amd dual core, what the intel Ghz equivalent roughly?
| « Programers of the world unite! | what to choose? B.Eng or B.Sc (Software) » |
| Bookmarks |
Bookmarks |
LinkBack URL
About LinkBacks
