- Joined
- Jul 30, 2001
- Location
- Mauritius
Thing is with the exception of the impossible to quantify computer performance none of the interpretations of Moore's Law have fit over the past 30 years.
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
Moore's Law Dead??!!!
- Thread starter syberspy9
- Start date
OP
- Joined
- Sep 4, 2003
- Location
- Salmon Arm BC CANADA
- Thread Starter
- #42
imo the law is the way he said it you cant change it or you would just be changing moore's law so that it would always be right i think by know it is dead as of today imo its dead by the time the presscot came out it was dead for shure its ben bent by to many ppl to many times i think we should move on and some on has to come up with a new theory
- Joined
- Mar 29, 2004
- Location
- California
Everyone keeps posting the "18 months" even though Moore never said this in his paper(s). At first he said "12 months" and then "24months" - and then all the others averaged it out.
It was also first related to the number of transistors per IC and NOT performance power (which is a relatively new twist by newszines). Yes, it has fallen out a bit but has held fairly well from 1960-2002 or so. Death may be imminent but we're not there yet.
Also note that in his paper Intel does not actually produce the maximum transistors per IC but a cost effective version - case in point ATI using 0.13u process has far more transistors than the P4 2.x B or P43.xC and more than the P4EE (half of the real estate on the die goes to cache!) ATI's latest has 220+million - unfortunately a large chunk of transistors are used for interconnects and signal buffering.
OC Detective: Great link!
You can see Intel's site:
http://www.intel.com/research/silicon/mooreslaw.htm
It was also first related to the number of transistors per IC and NOT performance power (which is a relatively new twist by newszines). Yes, it has fallen out a bit but has held fairly well from 1960-2002 or so. Death may be imminent but we're not there yet.
Also note that in his paper Intel does not actually produce the maximum transistors per IC but a cost effective version - case in point ATI using 0.13u process has far more transistors than the P4 2.x B or P43.xC and more than the P4EE (half of the real estate on the die goes to cache!) ATI's latest has 220+million - unfortunately a large chunk of transistors are used for interconnects and signal buffering.
OC Detective: Great link!
You can see Intel's site:
http://www.intel.com/research/silicon/mooreslaw.htm
- Joined
- Oct 12, 2001
- Location
- Folding in Barbados
Belephon ................ cool thanks .
- Joined
- Mar 29, 2004
- Location
- California
Cowbox X .... you're welcome.
CPL.Luke ... I'm not sure what you are trying to say, that is exactly what they are currently doing. Moore, Schokley (not the original inventor of the transistor - but the supervisor), and others observed that since the creation of the IC - all of our current silicon transistors work on the basic PN junction. (see previous posted article for summary.
Basically, all transistors have a Base, Collector and Emitter and in an IC or any semiconductor it uses Silicon (pure crystal grown, sliced, baked) since it is farily cheap and can be doped with other substances which converts it from a non-conductor to a conductor with the simple application of power (apply voltage).
So, you have a large sheet of silicon which is doped with different atoms of tantalum, gallium, boron, arsenic, etc, and where these impurities come together to create either a "P"ositive junction or an "N" type - they create a transistor ON the silicon. You then layer them to increase density and to provide a paths, power, ground, etc.
Ok... this is getting way too long winded and my memory of micro electonics is spotty, see the following:
http://science.howstuffworks.com/diode4.htm
http://science.howstuffworks.com/fr...chnik.uni-erlangen.de/lehre/mm/html/start.htm
As many have noted, and posted above... yes we are approaching the end. There is a real physical limit to how small you can create wire traces... a minimum 1 atom, actually more like 21 atoms as well as the very real problem of how to create photolith masks when the light being used has a longer wave length than the traces being created.
i.e. 65nm process is shorter than the standard UV light wave length - 2010 could be crunch time for one big hurdle.
ugh... must sleep.
CPL.Luke ... I'm not sure what you are trying to say, that is exactly what they are currently doing. Moore, Schokley (not the original inventor of the transistor - but the supervisor), and others observed that since the creation of the IC - all of our current silicon transistors work on the basic PN junction. (see previous posted article for summary.
Basically, all transistors have a Base, Collector and Emitter and in an IC or any semiconductor it uses Silicon (pure crystal grown, sliced, baked) since it is farily cheap and can be doped with other substances which converts it from a non-conductor to a conductor with the simple application of power (apply voltage).
So, you have a large sheet of silicon which is doped with different atoms of tantalum, gallium, boron, arsenic, etc, and where these impurities come together to create either a "P"ositive junction or an "N" type - they create a transistor ON the silicon. You then layer them to increase density and to provide a paths, power, ground, etc.
Ok... this is getting way too long winded and my memory of micro electonics is spotty, see the following:
http://science.howstuffworks.com/diode4.htm
http://science.howstuffworks.com/fr...chnik.uni-erlangen.de/lehre/mm/html/start.htm
As many have noted, and posted above... yes we are approaching the end. There is a real physical limit to how small you can create wire traces... a minimum 1 atom, actually more like 21 atoms as well as the very real problem of how to create photolith masks when the light being used has a longer wave length than the traces being created.
i.e. 65nm process is shorter than the standard UV light wave length - 2010 could be crunch time for one big hurdle.
ugh... must sleep.
OP
- Joined
- Sep 4, 2003
- Location
- Salmon Arm BC CANADA
- Thread Starter
- #47
well then they should make the cpu's a bit bigget i mean there small imo im not saying make the cpu the size of the p4 hs but mabey make them a bit bigger then later down the road they can wrk on cutting the size down like a tower cpu not a mini cube cpu or laptop cpu but if they made the next cpu a bit bigger but not thicker because if they did that then the it would be hott in the middle the would make it the same thicknes as the p4 but longer and wider by mabey 1cm more that would give them lots of space!!! and other than the size it would be worth it
i mean come on ppl if you had to make it 1 cm bigger for 2x the power of the 3.4ee would you??? i would i have lots of rome in my case
but it would be nice if when they did that the mobo companys made the new btx mobos rather than the old atx stile
http://www.intel.com/update/images/dt10031_g1.gif
good idea but they better make it so that hte vga card goes in the other way the core should be on top heat rises!!! ppl
i mean come on ppl if you had to make it 1 cm bigger for 2x the power of the 3.4ee would you??? i would i have lots of rome in my case
but it would be nice if when they did that the mobo companys made the new btx mobos rather than the old atx stile
http://www.intel.com/update/images/dt10031_g1.gif
good idea but they better make it so that hte vga card goes in the other way the core should be on top heat rises!!! ppl
