[mercury-users] 234-Trees vs alternatives in the compiler

[Ralph Becket]

Thomas Charles Conway wrote on 16 Dec:
> 
> I changed the runs to make them longer, because I was suspicious
> because the difference was so great. Here are some more reliable
> results on runs 10x as long:
> 	234 trees:		83.0 sec
> 	red black trees:	80.9 sec
> 
> which is a pretty small difference.

Sure, there shouldn't be much to choose between the two over 100,000
operations.  My question really assumed that N would be small in most
cases within the compiler.  In which case it might make sense to
experiment with, say, chained fixed-length arrays or somesuch.  Just a
thought.

Ralph

-- 
Ralph Becket  |  rwab1 at cam.sri.com  |  http://www.cam.sri.com/people/becket.html