Author: Vincent Diepeveen
Date: 16:28:33 01/19/05
Go up one level in this thread
On January 19, 2005 at 10:25:35, P L Patodia wrote: You must not confuse algorithmic enhancements with hardware enhancements. Moore's law as done in 1965: http://www.intel.com/research/silicon/mooreslaw.htm Until today Moore's law holds true. Based upon that we will not calculate 10^43 in or around 2066. See extensive calculations done by me a few years ago. I am however a generous man. If you claim we need 10^44 calculation power to make a database of size 10^43, i'll give you that 3 year extra. No problem. 2069. I'm a generous man. Please note that engineers must make ever more increasingly difficult inventions in order to keep up with Moore's law. I will not blame them if Moore's law ends within a few years from now. Big quantummechanica experts, such as Eeuwe Sieds Zijlstra, carefully warned me that the speed of light is 1/3C. In aluminium that gives 1.3Ghz at most. In copper a bit more. But there are ever more increasing limits to overcome. Just compare with this. The first aircraft which flew, whether it be a Frenchman or an US citizen, it was either a 1 men or a 2 men crew doing it. They did it on their own. The spaceshuttle required tens of thousands of engineers and designers to get flying. Hardware has that problem too. The unofficial second law of Moore is that with a 2 fold increase in transistors for newer processors, also the price of the factories to produce them doubles. A report from intel fears that a new factory of the future to produce processors will cost them $20 000 000 000 of investments. 20 billion dollar. That's quite some money. So therefore my statement that it won't be before 2066. >You cannot be sure of 2066 because prediction in Computers >and Chess are very difficult to make. > >Slight disgress: > >20 years ago, mathematicians were saying that we cannot factor >70 digit numbers. Today factoring 70 digits number is very >simply. The numbers of 129 digits (I am talking about general >numbers not with any specific properties) are broken by computer >scientists. > >How do you know what will happen tomorrow? > >On January 18, 2005 at 19:28:19, Vincent Diepeveen wrote: > >>On January 18, 2005 at 18:29:46, chandler yergin wrote: >> >>The good news is: it won't be solved before 2066, >>assuming that Moore's law applies also to increase of cpu speed and storage >>speed (which lately is not so sure anymore). >> >>The bad news is that chess has 10^43 possibilities (any no degree person can >>calculate for you it's less than 10^45, but some math guys calculated it to be >>less namely order of 10^43). >> >>In 1986 the XT could see say a few thousand nodes. (10^3) >>In 1996 we all could see hundreds of millions of nodes (10^8) >>In 2005 we can search after some days many tens of billions of nodes (10^11) >> >>Using Moore's law it is easy to see computers get faster each year. So they will >>also be able to see all chess positions, but 100% sure not before 2066. >> >>So the good news is that most likely we both won't see that date happen. In 2066 >>i'm 93 years old and my grandfather died at 88 and his father died at 88 too :) >> >>Knowing that i'm a FM, that's good news for both of us. >> >>Most likely however there is even better news than that. >> >>That's that there is physical limits and where no doubt each year they manage to >>get faster processors, the end is in sight. Electrons simply move at 1/3 of the >>lightspeed. No way to get them faster. >> >>Even lightcomputerchips just give a speedup of factor 3 over that, and for sure >>coming 20 years there won't be lightcomputers yet, such technologies eat way >>longer to become 100% reliable, let alone cheap. >> >>Harddisks do not grow bigger much now, though some argue that's because of >>price. >> >>We had 20MB in 1985 here or so at the businesscomputer. >>In 1995 i had a real poor machine so let's discount that, but it was like >>8 GB. >> >>Now 2005 i have several harddisks, but none of them is bigger than 200GB. >> >>I'm considering buying a raid5 controller now in order to generate 7 men in the >>attic (but i hardly have the money for a raid5 controller let alone 8-12 >>harddisks). >> >>Trivially the limits have been reached and it will be interesting how engineers >>manage to move on. >> >>So it is unlikely chess will be solved before 2066. >> >>Very unlikely. >> >>Note that Quantum computers do not help either to solve it. The only type of >>quantum i have a photo from is from the supermarket store around the corner here >>which is called quantum and they do not even sell computers, let alone quantum >>computers. >> >>Vincent
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.