|Date: ||Tue, 13 Nov 2001 22:20:16 -0500|
|Reply-To: ||Charles Patridge <Charles_S_Patridge@PRODIGY.NET>|
|Sender: ||"SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>|
|From: ||Charles Patridge <Charles_S_Patridge@PRODIGY.NET>|
|Subject: ||Re: Speediest text search?|
As Sigurd pointed out, you need to provide more info about your requirements
and how the data is structured for your searching.
I often have the same problem with my "fuzzy match/merge" routines but that
tends to be the nature of the beast. However, I do notice significant
savings of machine time depending on what OS I am using for the exact same
routines and data.
My best performance for medium size files and smaller is doing the work on a
stand-a-lone DeskTop with SAS loaded on it. Having lots of memory and 2nd
hard drive (spinning as fast as possible - 7200 rpm or more) helps a get
deal and out performs larger mainframes and midtier due to it (desktop)
being 100% dedicated to my job. But this may not be an option for you.
I have also seen some place (I think Paul Dorfman) was able to read certain
size files into memory and then do the searching but I do not know the
specifics. Maybe, Paul will see this and respond to you.
Overall, my experience is the routines become very CPU bound because of all
the searching for strings on a given record.
I do not think this helps you a great deal but I thought I would share a
little of my "fuzzy" thinking with you.