Author: Keith Evans
Date: 15:48:53 06/21/02
Go up one level in this thread
On June 21, 2002 at 14:45:57, Dann Corbit wrote: >On June 21, 2002 at 14:11:58, Randall Jouett wrote: > >>Howdy Dann, >> >>On June 20, 2002 at 15:04:46, Dann Corbit wrote: >> >>>On June 20, 2002 at 14:50:09, Robert Henry Durrett wrote: >>> >>>>On June 20, 2002 at 14:44:46, Osorio Meirelles wrote: >>>> >>>>> >>>>> How much more speed could we have than a PC chip, if there was a >>>>> hardware specifically designed to play chess ? >>>>> >>>>> Wouldn't this make an extreemly powerfull dedicated machine ? >>>>> >>>>> How much does it cost do develop such a Chip ? >>>> >>>> >>>>Multiprocessor, 64-bit [at least], and expensive. But, how much are REAL chess >>>>nuts willing to pay? >>> >>>That question is irrelevant. There are probably one or two people who would >>>play 1 million dollars for a machine that would beat Kasparov. But it would >>>cost 10 million to develop it. >>> >>>The real question is: >>>"How much are the broad masses willing to play for the world's strongest chess >>>machine?" >>> >>>The answer is "Not much." >>> >>>People balk at the cost of ChessMaster! CHESSMASTER -- for crying out loud. >>>They practically give it away. I have seen it online somewhere for $13. You >>>can walk into any computer department that sells software and get it for $30. >>>And people whine about that cost. >>> >>>You might sell computer boards with Hsu's chip on them for $2000, but to how >>>many people? I suspect that not one in ten CCC users would buy it, which means >>>that not one in one million of your average citizen would do so. >>> >>>You have to think about total cost of development compared to total return on >>>investment. >>> >>>That is the real problem and it is also the reason why we don't all have a copy >>>of the Hsu/Campbell chess machine buzzing away on our desk right now. >> >>I agree with all of these remarks 100%, which leads me to a question: >> >> >>Why haven't we seen a Beowulf-clustering program developed by a group >>of folks on the Intenet yet? I mean, I'm sure most of us are familiar >>with NetHack and the like, which was developed by various net users. >>Why not a NetChess that uses clustering? With a setup like this, >>it would seem that you'd probably gain 200 ELO points or something :^). >> >> >>Seriously, I'd imagine that companies such as Red Hat and the like >>would be more than happy to sponsor a setup like this in the WCCC. >>Mainly, I guess they'd pay for the phone time to connect to the >>cluster, because lugging a system like this to a tournament would >>be a major pain. OTOH, if each member of the NetChess team >>(which could be numerous) brought along a machine or two to the >>tournament, then maybe it wouldn't be all that bad. >> >> >>IMHO, a setup like this would totally dominate the WCCC. OTOH, >>what the heck do I know? :^) > >There are programs like that. >For intance: >Cilkchess Just a quick note - when I looked into Cilk a little while back it looked like they were more into SMP systems now (probably because they were more available to them and performed better) and weren't supporting AMP in their latest releases. I thought that Cilk was pretty slick though, and even ran a few programs on an SMP box running Linux. Unfortunately as far as I know they don't make the source to their chess program available. If I'm wrong about any of this, then please post details. >Star Socrates >Zugzwang >P.ConNerS > >Probably some others. Unfortunately, the speedup these programs get for having >a massive pile of CPU's is rather pathetic. For having hundreds of CPU's, they >have barely more compute power than a small SMP machine. > >I think someone needs to totally rethink the AMP chess paradigm and come up with >something better.
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.