For the use of fast networks:
yes, fast networks improve the results, in particular for 9x9, in my humble
opinion - however, you have already a good speed-up without
that, in particular for 19x19, and in particular if you have multiple cores
per node so that one core can take care of communications whereas the
other nodes compute simulations.

But for grids (instead of clusters), the communication will become much much
bigger - I'd like to study that
carefully one day, I have no clear idea of what is possible.

A trouble is that on grids (at least the ones I've seen) there are often
faults. We'll have to be fault tolerant I guess.

I point out that it has been said that very long delays between
communications are not a trouble - this is not the case
in my experiments, but perhaps this is implementation dependent, or perhaps
I am wrong :-)
Olivier
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

Reply via email to