@Tom wrote: All you did was move ID to the front since it was the only variable listed in your RETAIN statement.   Define something before the ACO variables and you will see that they do not "move". 13 data have; 14 input XX ACO_1-ACO_3 ID; 15 put (_all_) (=); 16 datalines; XX=5 ACO_1=11 ACO_2=111 ACO_3=1111 ID=1 NOTE: The data set WORK.HAVE has 1 observations and 5 variables. NOTE: DATA statement used (Total process time): real time 0.02 seconds cpu time 0.00 seconds 18 ; 19 20 data test; 21 retain ID ACO_: ; 22 set have; 23 put (_all_) (=); 24 run; ID=1 XX=5 ACO_1=11 ACO_2=111 ACO_3=1111 NOTE: There were 1 observations read from the data set WORK.HAVE. NOTE: The data set WORK.TEST has 1 observations and 5 variables. NOTE: DATA statement used (Total process time): real time 0.00 seconds cpu time 0.00 seconds Thanks @Tom   ... View more

Please post the complete(!) output of proc contents for all three datasets. ... View more

@hanfei28 wrote: Thanks for your answer @Kurt_Bremser @Reeza    Here are my system information. Viya system has one machine with 16 CPU cores, RAM=126GB.   As long as you don't run it in CAS, the size of your WORK (on disk) is important. Have that checked. After switching to CAS, since the projected size of the 1 million multiple dataset is 66 GB, that would eat more than half of your available RAM. As an admin, I would NOT let a single user occupy half my space.   Bottom line: reduce the multiply factor to a value that is sufficient to display the performance difference between SAS 9.4 and CAS/Viya. ... View more