just wanted to give a quick example using a relatively small dataset: NOTE: The data set WORK.TEMP has 453310 observations and 157 variables. 28 proc sql; 28 ! create table temp2 as select 29 * 30 from temp(where=(current_balance>5000)); NOTE: Table WORK.TEMP2 created, with 419030 rows and 157 columns. 31 quit; NOTE: PROCEDURE SQL used (Total process time): real time 2.39 seconds cpu time 2.39 seconds 32 33 proc sql; 33 ! create table temp2 as select 34 * 35 from temp 36 where current_balance>5000; NOTE: Table WORK.TEMP2 created, with 419030 rows and 157 columns. 37 quit; NOTE: PROCEDURE SQL used (Total process time): real time 3.20 seconds cpu time 2.54 seconds as you can see above the real time processing is almost a second quicker when processing at compile time and if you scale that up for big data, multiple users, multiple data sets it quickly becomes a big difference
... View more