[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: computer-go: Computer Go hardware
Nick@xxxxxxxxxxxxxxxxx (Nick Wedd) wrote:
>You have not defined an index. You have labelled two points on an
>arbitrary scale.
>Here is a defined index: the number of points a program gives away in
>the course of a game, by its inferior moves. For the best current
>programs, this is around 150. For perfect play, it is 0.
I agree with you. What I wrote was trying to give an estimate how efficient
the best programs are using the available computing power. Someone may say
these programs only begin to touch the tip of an iceberg. I don't think so.
There is a randomness in each game. Only average over many game can give
accurate results. Also if program only compare with program, it's still a
relative number.
heikki@xxxxxxxxxxxxxxxxx wrote:
>.............
I agree with most what you said, except you neglected one detail. It takes
time to do game planning etc. At present programs used up most of the time
during a competition. To implement what you said in addtion more computing
time is needed. It's not a simple algorithm to implement what you said. You
mentioned higher level abstraction of the game. It's exactly this higher
level abstraction that made it diffcult to implement an efficient algorithm
in FPGAs. This higher level abstraction takes more time than the lower level
abstraction.
Dan Liu