I notice that the 2008 icga chess tournament is limited to 8 cores.
David Levy's justification seems curious to me. He mentions that an early
microcomputer held its own against a mighty mainframe, and that many top chess
programs run on PCs, but he wishes to discourage being able to "buy the title"
by using larger clusters of computers.
The 8-core limit seems like the constraints on some car races, which limit
displacement, intake sizes, and a host of other variables in order to make the
race more exciting. If the cars are all the same, then driver skill is what
really matters.
But computers have a lot more flexibility. Should these be eight AMD cores or
Intel cores? x86 or Itanium? or PowerPC? Should they be on a single 8-socket
motherboard, or four or eight motherboards tied together by ethernet or
infiniband? How about FPGAs with hundreds (or even thousands ) of tiny
special-purpose processors? What about overclocking? There are folks who claim
6 or 8 GHz speeds with nitrogen cooling.
There's a place for competitions with the same resources across the board, but
there's also a lot of excitement in the "run what you brung" competition, of a
different sort. This is all the more true when - unlike Formula 500 cars - the
average enthusiast can reasonably expect to drive something like the current
supercomputers in a few years. Something like the 80-core Larrabee might be on
our desktops in a decade.
I think that lower-budget teams with "only" eight cores will be driven to
explore approaches which use the hardware more optimally.
Terry McIntyre <terrymcint...@yahoo.com>
-- Libertarians Do It With Consent!
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
