Date: Sat, 5 Jul 2008 08:32:32 -0700
Reply-To: KL <Kev.L.Scott@GMAIL.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: KL <Kev.L.Scott@GMAIL.COM>
Subject: Re: PROC SORT
Content-Type: text/plain; charset=ISO-8859-1
On Jul 5, 10:36 am, r...@CDC.GOV ("Fehd, Ronald J. (CDC/CCHIS/NCPHI)")
> > From: SAM
> > I wanna know if there is a way to sort data more fast than the "PROC
> SORT" do.
> what are your priorities?
> * accurate
> * cheap
> * fast
> as always, your two top priorities determine which algorithm you
> for your Reading and Research
> 1. use this code to examine other options related to sorting
> 2. RTFM on and the sorting options found above
> 3. consider
> * buying SyncSort
> * indexing, see Raithel's book as mentioned in earlier post
> Ron Fehd the macro maven CDC Atlanta GA USA RJF2 at cdc dot gov
We also need to know what the programmatic context in which the SORT
is being performed, please refer to my questions regarding why the
data needs to be sorted.
For example If it is for merging/joining purposes then the sort can be
avoided either by using a hash based join in data step or SQL or a SET
KEY= index join if the amount of records being joined in the master
table is small relative to its overall size.