Date: Fri, 19 Apr 2002 12:43:19 -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: Hardware questions
Content-type: text/plain; charset=us-ascii
Joe McCrary <joe.mccrary@GADATA.ORG> wrote [in part]:
> Right now, we are using PC SAS, 8.12, on Win 2000 machines. About 512
MB Ram, 2 20-GB HD's.
> P4 1.4 GHz. I know for a fact we'll need to upgrade our hard drives to
several hundred MB's.
> My questions are this:
> 1. Would upgrading the amount of RAM make sense? Would I see a
tremendous performance increase?
> 2. Would a faster processor make a tremendous difference?
> 3. Are there other solutions we ought to be exploring (non-SAS, such
as SQL-Server or Oracle)?
> My inclination here is no, because we need the multivariate analysis
capacity.
[1] RAM can help, but given the capabilities of the upcoming SAS V9,
you might seriously consider investing in a 2000/XP box with multiple
processors. V9 will allow you to parallelize your sorts and proc means
and so on, as well as pipelining your data steps. If you do up your
RAM,
consider putting in a *lot* of RAM. The SASFILE statement will let you
load a file into memory for faster access, and being able to put all
your
working file in RAM could speed up your work considerably, depending on
your usage. You might also want to look up some papers from this and
earlier
SUGIs on dealing with large datasets, sorting vs. indexing, and similar
topics which are causing you grief.
[2] As above, multiple processors may help you more than a faster
processor.
[3] SQL-Server and Oracle are great at transaction processing, but they
just don't compare to SAS on read access for data manipulation or data
analysis. They are tuned with different goals in mind, so why should
they?
Paul Dorfman has made at least one comparison in this list in the past
year,
so you could look up his post off the SAS-L archives:
http://www.listserv.uga.edu/archives/sas-l.html
Some posters in this group find that using Google to look up past posts
is very fast.
HTH,
David
--
David Cassell, CSC
Cassell.David@epa.gov
Senior computing specialist
mathematical statistician
| 
