| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.server -> Re: SQLServer beats 8i finding top-10 duplicates
Jonathan Lewis wrote:
>
> Damn !
>
> A technical term I don't understand.
>
> Anyway I suspect you gave Oracle a
> worst case scenario - I dropped my
> sort_area_size to 4M, and retained to 3M
> and still got a pretty good result before
> going to disc - a much more reasonable
> memory demand - and probably what
> MS Server was doing anyway.
>
> I used the dbms_random function 1,000,000
> times to get my data (20 minutes CPU !).
> It certainly seemed to look nicely random.
>
> Is LC the one where you keep multiplying
> a couple of big numbers and taking bits
> out of the middle ?
>
> --
>
> Jonathan Lewis
> Yet another Oracle-related web site: http://www.jlcomp.demon.co.uk
>
> Connor McDonald wrote in message <3879D379.3FE5_at_yahoo.com>...
> >
> >I used a linear congruential - I didn't check its period, but as with
> >all LC generators, it would have cycled after some amount of time...
> >
> >A nice healthy sort_area_size = 16M certainly assisted in the timing
> >results. I suppose you could call that cheating a little. :-)
> >
I'm back in Win98 now so I can't see the generator source at the moment, but linear congruential generators are all of the type:
S(x+1) = aS(x) + c modulo M
I generally use them because certain types of them are great for generating "random" but non-repeating numbers which is handy when I'm doing tests for "random access" to a table.
A nice reference:
www.ulib.orig/webRoot/Books/Numerical_Recipes/bookcpdf/c7-1.pdf
Cheers
C.
--
"Some days you're the pigeon, and some days you're the statue." Received on Tue Jan 11 2000 - 07:21:22 CST
 |
 |