On Jan 24, 2008 8:00 AM, Heikki Levanto <[EMAIL PROTECTED]> wrote: > Everyone is looking at the top end of the scalability study > http://cgos.boardspace.net/study/ > > But what happens in the low end? Both programs show linear progress to > begin > with, then a corner, and more (almost?) linear development.
I think you're looking at ghosts in the data. What if the fatman's point for 3 was slightly lower (well within the error bars that aren't shown). That then would make the "corner" completely disappear. You'd probably even say it was near-constant rate of growth the whole way up (maybe a gradual roll-off). > Fatman's curve has a clear break at 3 doublings, when it suddenly starts > to > improve much slower than before. This goes on until 12 doublings, after > which > we get the mysterious decline. The decline can simply be a fluke. More data is needed before any final conclusion can be reached. I'm actually surprised at how smooth the rest of the data is. I think our eyes are likely playing tricks on us. It's better to hold off on subtle interpretations until you have enough data to discount the effects of noisy data points.
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
