Olivier Teytaud wrote: >> Would you guess that mogo is 2 or 3 ranks stronger at 19x19 with all >> this hardware? > > I just claim that mpi-mogo wins with very high probability against > sequential-mogo in 19x19. But I'm afraid that the improvement is > disappointing against humans. Hopefully it is still much stronger in reality. But I think your opponent in this case is too strong to really get much of a sense of what is happening.
If Mogo really isn't improving against humans but is improving against other Mogo's and this is a substantial effect, it means something is wrong with the algorithm - I would guess this would indicate that Mogo is too selective. I feel your pain - there is no easy way to test any of this without more power yet. If you need a network of workstations to test a single processor program, then you need a several networks of workstations to test a single network of workstations! - Don > > I hope better improvements are possible thanks to the fact that > parallelization makes heavier playouts computationnally less expensive > - communication and concurrency for memory access becomes negligible > with heavy playouts. > > _______________________________________________ > 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/
