How high can CPUs go?

[KillaKilla]

Disregarding cooling, would CPU's theoretically be able to reach unbelievably high speeds(like 30Ghz)? Or is there some kind of lock on the CPU (not the chipset or the MB) that prevents extreme overclocking?

 

[BlueWeasel]

Moore's Law??

 

[Falloutboy]

in theory yes but cpu's are hitting a wall today because its getting harder and harder to shirnk down to the next manufacturing size. I think the new thing in the next few years will be chips with 2-4 cores intergrated right on the die as opposed to xtreme ghz

 

[DAPUNISHER]

Originally posted by: Falloutboy525
in theory yes but cpu's are hitting a wall today because its getting harder and harder to shirnk down to the next manufacturing size. I think the new thing in the next few years will be chips with 2-4 cores intergrated right on the die as opposed to xtreme ghz

Multiple physical processors witH HT seems like the future of the Intel desktop line-up . Duvie is building a Xeon dually so I'm very interested to see how 2 CPUs with HT do with his encoding and CAD/CAM needs.

 

[preslove]

AMD to launch dual-core chips in 2005

Advanced Micro Devices Inc. will release dual-core Opteron processors for servers and workstations in 2005, the same time frame that Intel Corp. plans for its first dual-core products, an AMD executive said Monday...

Several chip makers are looking to dual-core designs as a way to increase the performance of next-generation processors without the constraints imposed by rising levels of power consumption in single-core processors...

IBM Corp. and Sun Microsystems Inc. have already released dual-core chips for high-end servers. The server world is especially primed to take advantage of multicore technology because so many applications for servers are already multithreaded, Williams said...

Earlier this month, Intel announced that it would shift all of its future server, desktop, and notebook processor development projects to dual-core designs. By 2005 Intel expects to have a dual-core chip available for each of those three segments, said Intel President and Chief Operating Officer Paul Otellini last week during the company's biannual financial analyst meeting in New York.

 

[clarkey01]

IBM showed off a 10Ghz on a 0.13 micron process a while back

http://www.theregister.co.uk/2000/12/11/ibm_reveals_10ghz1/

The Intel chip code named Nehalem was suppose to reach 10.25 Ghz In 2005, lol..

 

[clarkey01]

ah, its not a chip, its the tech, whoops, sorry

 

[jagec]

Originally posted by: BlueWeasel
Moore's Law??

Moore's Law != a law.

 

[Soulkeeper]

we could probably produce a 20ghz chip with 10x the performance of current chips but the cost to get the equipment/expenses/time/effort to do such a thing would be in the 10's of billions if i had to guess

anyways what's the use of toping out the theoretical limits of a current technology if other interconnected/dependent techs like memory, vid cards, hard drives, internet connection tech, etc aren't also updated in bandwidth/capacity/latency/cost
which when all is said and done is 100's of billions

just give them time, when certain researcher's or insiders say 20ghz or 30ghz is the limit they usually mean that's how high we know we can/will be able to go if cost/time/testing were all all all infinite.
atleast this is how i kinda see it.

but the industry and all it's technologies based off of silicon/electrical, current like cpu's, will only increase with time/demand/investments into upgrades design/cost price reductions.

moore's law has seemed to serve more as a standard set by a person/group of thousands to uphold, but breakthroughs and advancements in cpu speed could be considered like the stock market; the speed/power/effiency/etc balance seems to only improve with the $$$$ to back it up

 

[chsh1ca]

When discussing limits I think there are none. Ten years ago 1GHz seemed an impossibility. Right now 1THz seems an impossibility. Give it 10 years. Look at how far everything has come: Drives in the last ten years have gone from being 250MB to being 250GB, Internet access has gone from 19.2kbps to 3mbps/192kbps (at least in the areas I know of) for broadband, some areas higher. Researchers have found ways to increase that to 50mbps over regular copper-wire systems. Graphics cards have gone from processing simple 2D images to fully rendering hundreds to thousands of models and effects in realtime. If we've come this far in 10 years, do you not think we will figure out ways to continue the trend? We've managed to stop light, which could lead to new communications and/or interconnection technologies. We have the research done on how to convert a single electron into a transistor. With all these technologies being current, indeed it will take them some years to reach mainstream marketability and useability. Think about it this way though, we have all the intervening technologies to go through before we get to cutting edge for 2004 knowledge. At the same time, more and newer research will be done, new things developed, etc., etc..

The short answer to our question is: The wall will first come to be when all of our potential needs are sated, once our needs and wants are met, more than that level of technology would be superfluous.