Okay, here are some more numbers. I ran my test three times today (sorry, I
got some "real" work to do!). Here are the results:
Test program with 'debugging mode' code in it, but no 'PARM=1':
Run #1: 406.089 CPU Seconds
Run #2: 393.254 CPU Seconds
Run #3: 398.965 CPU Seconds
Total: 1198.308 CPU Seconds, Average: 399.436 CPU Seconds
Test Program without 'debugging mode' code:
Run #1: 361.347 CPU Seconds
Run #2: 376.105 CPU Seconds
Run #3: 370.126 CPU Seconds
Total: 1107.578 CPU Seconds, Averag: 369.193 CPU Seconds
Comparison: 399.436 CPU Seconds versus 369.193 CPU Seconds, 30.342 CPU
Seconds or 8.2% more CPU with 'debugging mode' code.
Test parameters: Program that reads over 600,000 records from an Image
detail data set using serial read. Code is identical with the exception of
the 'debugging mode' clause on the source-computer and the declaratives that
display the debug stuff. Run #1 was performed serially (i.e., one after the
other), while runs #2 and #3 were run in batch in 'parallel'; i.e., at the
same time.
Just for fun, I ran the program with the 'debugging mode' code in it using
'PARM=1' and it consumed 721.586 CPU Seconds, or almost twice that of the
program without the 'debugging mode' code in it.
Now, that was a 'batch' type program. We have a complete online system with
the 'debugging mode' clause on every one of the subprograms. I have no idea
if the overhead is as bad in the online environment as well (but, one would
assume so). Probably more testing is needed, it's just that the online
environment doesn't lend itself to that type of stuff, but we shall see....
Jim Phillips Manager of Information Systems
E-Mail: [log in to unmask] Therm-O-Link, Inc.
Phone: (330) 527-2124 P. O. Box 285
Fax: (330) 527-2123 10513 Freedom Street
Web: http://www.tolwire.com Garrettsville, Ohio 44231
----- Original Message -----
From: David Burney <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, August 21, 2000 10:41 AM
Subject: [HP3000-L] 'with debugging mode' overhead
> Greetings listmembers and listmemberettes.
>
> I've a question concerning the use of 'with debugging
> mode.' in Cobol code. I use this occasionally in my
> code to determine logic flow and variable values at
> specific points in the code. Yet, I do not use the
> "D" indicator in column 7. My question is, when I
> compile the code and run the object without using the
> ';parm=1' parameter to invoke the debug process, is
> there any performance loss ? In other words, is leaving
> the 'with debugging mode' code in the program, even
> though I don't invoke it with ';parm=1' more system
> intensive ?
>
> As always, TIA.
>
> --------------------------------------------------------------------------
--
> -------
> David Burney dburney @
> summitracing.com
> Summit Racing Equipment http://www.summitracing.com
> -----------
> "There's no such thing as a bad day when there's a doorknob on the
> the inside of the door." - Lieutenant Paul Galanti
P.O.W.
> -----------
> All opinions expressed herein are my own and reflect,
> in no way, those of my employer.
|
