Re: Performance test: DOW vs FOR.... interesting results...

Hey Mark,
I am on V5R4 compiling for V5R4.

The code is at the code400 link I provided earlier in the OP.

But here is a code.midrange.com link....
http://code.midrange.com/e651e483d6.html

I think I grabbed everything needed to pretty much just compile it.


Thanks
Bryce Martin
Programmer/Analyst I
570-546-4777



"Mark S. Waterbury" <mark.s.waterbury@xxxxxxxxxxxxx>
Sent by: rpg400-l-bounces@xxxxxxxxxxxx
06/22/2011 11:46 PM
Please respond to
RPG programming on the IBM i / System i <rpg400-l@xxxxxxxxxxxx>


To
RPG programming on the IBM i / System i <rpg400-l@xxxxxxxxxxxx>
cc

Subject
Re: Performance test: DOW vs FOR.... interesting results...






Hi, Bryce:

Some questions:

1. what version / release of OS/400 or i5/OS are you compiling and
running this on?
2. what TGTRLS are you specifying on the CRTRPGMOD or CRTBNDRPG
commands?

Also, could you post (a small version of) your "test" program(s) at
code.midrange.com and post the link generated, so that some of us who
might want to try to reproduce your results, on different versions /
releases of OS/400, can do so more easily?

Thanks,

Mark S. Waterbury

On 6/17/2011 9:37 AM, Bryce Martin wrote:
Ok, so out of curiosity yesterday I was doing some DOW vs FOR

comparisons

and thought that while I was at it I would do a performance test...and
this uncovered something interesting...

Here's the long of the short of it...
DOW is 3x faster than FOR. If you use OPTIMIZE(*FULL) then the gap is
even greater.

Now, before anyone says it... yes...we're talking about milliseconds

here.

But think of all your programs that have loops...yep, almost all of

them.

So I guess my main question is this...

Why the big relative difference? How is it that FOR can be so much

worse

than DOW? You can actually see a difference if doing 10's or 100's of
millions of iterations. So while most people won't see a difference in
individual instances, if all your FOR loops were 3 times faster in all
your programs, I can't help but think you might be able to see some

actual

performance across a fairly taxed system.

I don't know... just ruminating. Anybody have any gritty details?

See this link for code and discussion....

http://www.code400.com/forum/showthread.php/10499-DOW-x-lt-y-is-3-times-faster-than-using-for-x-number-downto-1

Thanks
Bryce Martin
Programmer/Analyst I
570-546-4777
--- This message (including any attachments) is intended only for the

use of the individual or entity to which it is addressed and may contain
information that is non-public, proprietary, privileged, confidential, and
exempt from disclosure under applicable law. If you are not the intended
recipient, you are hereby notified that any use, dissemination,
distribution, or copying of this communication is strictly prohibited. If
you have received this communication in error, please notify us and
destroy this message immediately. ---