Re: Nearest Common Ancestor Report (XDb1's $1000 Challenge)

> Note: the output shows the same starting and ending date/time. This
> indicates an elapsed time of less than 3 milliseconds (the smallest unit
> of time CURRENT_TIMESTAMP can measure). System used: PC running Windows
> 2000 + MS SQL Server 2000, AMD Athlon 1.3GHz, 256MB RAM.

Among other things, the difference in normalization between the implementations (still looking it over) is quite different. Based on a cursory look, it seems your implementation is simply utilizing the hierarchy table and not having to resolve any keys to generate the report (which was not the desired intent). In an initial attempt to make them more comparable, I modified XDb1 algorithm so as to not resolve things' names and simply prints their IDs. Based on a few runs, the report generation time is 2 or 3 ms using the hi-resolution QueryPerformanceCounter function provided by NT on a 500 Mhz PC. In general, the execution is linear to PC's speed thus 2 or 3 x (500/1,300) should be approx 0.77 to 1.2 ms on a 1.3Gz PC. These numbers are too small to make reasonable comparisons. Received on Tue May 18 2004 - 09:23:06 CEST