05-23-2006 10:42 PM
I have two programs that obtain the same result and expend the same
time but using different techniques: one uses a lot of access to the bseg table with primary index from an
internal table with bkpf information) and the other uses more abap processing ( using the internal table with bkpf
information and an internal table with bseg information, I use parallel
index).
When I analyse both programs with SE30 transaction the first program
expend 88,333,698 microseg. and the second expend 345,677 microseg. I
don´t understand this result because both programs expend the same time.
And the other issue is that I ran again the analyse transaction for the
second program and I obtained similar results with the first program.
05-23-2006 11:07 PM
I think this may be due to some buffering issues. When you run two programs back to back that access the same data, the one that runs first generally takes longer. Try running each a number of times and take the smallest time for each program.
I liked the blog Saquib.
Rob
05-23-2006 10:53 PM
Check Blogs !!
/people/rob.burbank/blog/2006/02/07/performance-of-nested-loops
Hope thisll give you idea!!
<b>P.S award the points.</b>
Good luck
Thanks
Saquib Khan
"Some are wise and some are otherwise"
05-23-2006 11:07 PM
Hi Saquib.
Sorry but this doesn't answer my concern.
I don't understand the high difference in microseg comparing the SE30 results but when I normally execute both programs, they consume the same time.
05-23-2006 11:13 PM
It depends on lot of things , can you post your code for both program .
It also depend how your system performance parameter setup as "Rob" already mentioned.
Let us execute on our system and see how they behave.
Thanks
SK
05-23-2006 11:43 PM
Thanks Rob and Saquib.
Sorry I can't post the code for these programs because are very huge (2000 approximately).
05-24-2006 12:17 AM
the micro secs is the actual amount of cpu time used, in your small example the elapsed time may be similar but one program actually used a lot more of the system resources than the other. If there were no other users on and no other processes running them the elapsed times would probably be different with the more efficient one being quicker but r/3 is unlikely to often give you that opportunity.
05-23-2006 11:07 PM
I think this may be due to some buffering issues. When you run two programs back to back that access the same data, the one that runs first generally takes longer. Try running each a number of times and take the smallest time for each program.
I liked the blog Saquib.
Rob