Keep dataset option. Reading large tables.

[Opa4ki]

Hi.

Will be there or not any difference in perfomance in following situation.

I have two tables.
Table A has three columns and 10G in size. I need data from two columns 'a' and 'b'.

Table B has 50 columns and 100G(or more) in size. It has the same 'a' and 'b' columns and same data in them.

If i'll use code like this

data SOMETHING;
set A(keep=a b);
something;

data SOMETHING;
set B(keep=a b);
something;

Would it be any difference in performance?
Or it doesn't matter which table I'll read - Big or Small. Message was edited by: Opa4ki

[milts]

Hi,

You might want to test it yourself by adding the fullstimer option. Then try to compare the results.

Best Regards,
Milton

[Patrick]

I would expect that reading table A performes better as there should be less i/o.
As mentioned already fullstimer would give you some real values.

There was lately a interesting discussion in the SAS newsgroup which also might give you some insight:
http://groups.google.com.au/group/comp.soft-sys.sas/browse_thread/thread/7cc579c99df3badc?hl=en