|Date: ||Tue, 15 Dec 1998 09:01:02 -0600|
|Reply-To: ||"McKown, John" <JMckown@INSURDATA.COM>|
|Sender: ||"SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>|
|From: ||"McKown, John" <JMckown@INSURDATA.COM>|
|Subject: ||Re: fastest way to sort a large data?|
Forgive me for second guessing you. But could you possibly get away with
using a INDEX instead of sorting the dataset? If this is a permenant SAS
file that gets observations added to it, this should be faster than
resorting the entire dataset after adding observations.
Hope this is of some use,
> -----Original Message-----
> From: Sung-Il Cho [SMTP:sungil@HOHP.HARVARD.EDU]
> Sent: Tuesday, December 15, 1998 8:39 AM
> To: SAS-L@VM.MARIST.EDU
> Subject: Q: fastest way to sort a large data?
> Dear list,
> I wonder what is the fastest way to sort a large data set?
> Unfortunately I need to sort a data set of 30,000-40,000 records
> many times (say 100 times a day) for some things involved in simulation.
> There are two BY variables, e.g. subject ID and date.
> I know it would sound absolutely crazy, but assuming that I really
> need those sorting, what would be the fastest way to do it?
> So far I've been just doing PROC SORT, which seems to make my
> program run for hours.
> I guess I've heard about some alternatives
> to PROC SORT, but I have no idea what those are.
> I'm using SAS v6.12 on Unix with SunOs.
> Thanks very much!!