Also you have to look at the amount of computation per unit communication. For instance factoring huge numbers is an excellent application for a grid because
with just a few bytes of data, you can keep a CPU busy for a very long time. Regular Go programs don't fit this model. Maybe there something you can do to
help the problem. In an extreme case, one could distribute a single Go position to each machine and ask whether or not the position is a win for black. In
that case, you have met the original constraint (low communication, high computation) but it is obviously not practical. It's just an extreme example.
David Doshay wrote:
The @home systems work great for big problems that do not have time
constraints. Game playing is interactive and people expect reasonably
quick replies. The problem with @home computational models is that you
never know when the user will want their machine back, so you have the
problem of deciding if a result is worth waiting for, or if you will
send similar requests to multiple machines just to try and be sure that
you get at least one reply.
I was contacted by someone in the Govt of Singapore about trying exactly
this ([EMAIL PROTECTED]) and while it is interesting, it is not nearly as simple
as SETI@ home, where independent problems are being solved on different
machines and you do not expect to get the answer back to the primary
server on any particular schedule. In Go the answers are interdependent.
I suggested that it was too hard, or at best a research project not
nearly ready for prime-time.
Cheers,
David
On 2, Oct 2008, at 10:17 AM, Michael Markefka wrote:
So, when are we going to see distributed computing? [EMAIL PROTECTED],
[EMAIL PROTECTED], [EMAIL PROTECTED] With Go engines that scale well to increased
processing capacity, imagine facilitating a few thousand PCs to do the
computing. For good measure, [EMAIL PROTECTED] as about 800,000 nodes online
as of now.
What's the approximate increase in playing level per increase in
processing power? Any rough law for that?
Best regards,
Mike
Olivier Teytaud wrote:
Mogo was allowed to use 800 cores, not more, and only for games
against humans.
We have no acces to so many cores for computer-computer games (if
there were only three teams involved,
we could :-) ).
For some games Huygens was unaivalable at all, and mogo played with
much weaker hardware (some quad-cores,
however, it is not so bad :-) ).
Best regards,
Olivier
------------------------------------------------------------------------
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
