Ieronim,
I have finished benchmarking the routines. I didn't include any from
this thread or
this one but you are welcome to. I did not penalize any routine for processing the dictionary. Additionally, if there were a number of small variations for a routine, I only included the fastest.
Rate limbic__2 solo____1 limbic__3 limbic__4 ieronim_1 lim
+bic__1
limbic__2 6.48/s -- -45% -62% -90% -95%
+ -96%
solo____1 11.8/s 82% -- -31% -82% -92%
+ -92%
limbic__3 17.3/s 166% 46% -- -74% -88%
+ -89%
limbic__4 67.5/s 941% 471% 291% -- -53%
+ -55%
ieronim_1 143/s 2104% 1109% 728% 112% --
+ -5%
limbic__1 150/s 2218% 1171% 771% 123% 5%
+ --
Please note that the winner is just my re-write of
Ieronim's code with a few extra bells and whistles. Since the benchmark is extremely long, I have put it in spoiler tags as well as readmore tags:
Update (2006-07-03): These benchmarks do not reflect the realization I made concerning when to test if a path connection has been made that I noted elsewhere in this thread.