When I want to compare two or more methods, what is the important?

[AndersS]

When I want to compare two or more methods, what is the important?

Shall I use USER CPU TIME  or  SYSTEM CPU TIME   ?

Shall I use Memory   OR   OS memory?

See the text below (provided by options FULLSTIMER)

My method will not use a lot of memory – because I do not use sort very much.

Cpu time for My method – I do not know quite yet.

My guess:  USER CPU Time and perhaps Memory. See below.

       real time           0.02 seconds

       user cpu time       0.00 seconds     ß

       system cpu time     0.01 seconds ß

       memory              7285.46k             ß

       OS Memory           35788.00k       ß

       Timestamp           2022-05-06 03:51:12 em

       Step Count                        478  Switch Count  4

       Page Faults                       0

       Page Reclaims                     1676

       Page Swaps                        0

       Voluntary Context Switches        70

       Involuntary Context Switches      0

       Block Input Operations            0

       Block Output Operations           264

[SASKiwi · Accepted solution]

What is important to you regarding SAS performance statistics is entirely up to you.

If a SAS job is taking hours to run and you want it to run quicker then real time (elapsed time) and CPU time will be important. If a SAS job fails because it has run out of memory then the Memory usage will be important.

[AndersS]

Hi! USER CPU time and USER memory are the two measures, in classical reporting.
Of course Accuracy, not failing are important.
There does not seen to be a straight forward way of comparing two methods.

[AndersS]

Hi!

I have now been using CARY SAS CAS for some computing.

Result:  In my opinion BOTH "User CPU time" and "System CPU time" must be used, when you want to add the resources used for a method.

user cpu time 1.23 seconds

system cpu time 4.56 seconds

To only use "User CPU time" will underestimate the total amount of resources used.
/Br Anders