Date:         Mon, 26 May 2003 14:41:43 -0700
Reply-To:     cassell.david@EPAMAIL.EPA.GOV
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "David L. Cassell" <cassell.david@EPAMAIL.EPA.GOV>
Subject:      Re: QuickSort algorithm implementation in PC environment for
              Massive datasets
Comments: To: "Kundu, Om (US - Boston)" <omkundu@DELOITTE.COM>
Content-type: text/plain; charset=us-ascii

"Kundu, Om (US - Boston)" <omkundu@DELOITTE.COM> wrote:
> Leads on open-source code out there for the best SAS implementation of
the
> QuickSort algorithm (or something even more efficient!) for sorting
obs in a
> dataset for the PC environment that scales well for massive datasets
would
> be appreciated.

You seem to be under the misapprehension that the 'quicksort' is
quick.  I.e., that quicksort is the fastest thing you can do.  That
simply is not true.  There is no reason not to use the SAS heapsort
process on your PC, unless you have specific data which make one of
the O(n) techniques [radix sort, for example] feasible.  In particular,
if you have SAS v9, you can benefit significantly from the heapsort
implementation on a PC with multiple processors.

Alternatively, go with a mainframe and SyncSort.  That's going to
be the case where a sort works significantly better than the SAS
internal sort.

HTH,
David
--
David Cassell, CSC
Cassell.David@epa.gov
Senior computing specialist
mathematical statistician