From: Jonathan Franklin (franklin.jonathan_at_[hidden])
Date: 2008-03-05 15:34:36
On Wed, Mar 5, 2008 at 1:14 PM, Michael Fawcett <michael.fawcett_at_[hidden]>
wrote:
> Same test, but I resized the image to 1280 x 1280 (it was at 128 x
> 128). There were 1,016,906 nodes generated during the search.
<snip>
> at_goal() - 19.05272 seconds
> exception - 18.85408 seconds
So on this platform/configuration, w/ ~15K nodes, at_goal() was ~7.5%
faster, and w/ ~1M nodes, exception was a little over 1% faster. It would
be interesting to plot some curves to see where the turning point for this
platform lies.
I still wonder if the exception method would *significantly* outperform
at_goal() for larger graphs. For me, 1% isn't all that compelling, unless
it offers something else, such as design simplicity (or decreasing the
WTF?!? (astonishment) factor).
Jon