|
View previous topic :: View next topic
|
| Author |
Message |
CICS fan
New User
Joined: 03 Apr 2008
Posts: 82
Location: United States
|
|
|
|
Hello,
We have upgraded to enterprise PL/I and this has caused one of your PL/I programs to increse its run time by 4 times. There has not been any change to the code.
Anyone here has faced this or can please suggest a solution?We tried using compiler parms like OPTIMIZE(TIME) etc but to no avai.
Thanks in advance. |
|
| Back to top |
|
 |
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8696
Location: Dubuque, Iowa, USA
|
|
|
|
From the Compiler and Run-Time Migration Guide:
| Quote: |
1.2.5 Major changes with the new compiler and run-time
With Enterprise PL/I, you will find that existing PL/I applications are affected by several areas such as removed or changed compiler options, different default compiler options, and restrictions in combining old and new load modules.
The following list of concerns is merely a representative list that reflects what has been important to some customers. It may not indicate what is important to any one individual customer. More details are provided in the rest of this book.
Enterprise PL/I supports only Language Environment releases currently in support.
Enterprise PL/I has no support for VM.
Enterprise PL/I has no support for multitasking (but it does support multithreading).
Code that is incorrect or invalid (for instance, code that uses uninitialized variables) may not run the same. This may not seem like an important problem, but it has been a significant issue for most of the customers that have migrated.
You may need to specify some non-default options to get the most compatible behavior from the compiler and to get the best performance from the compiler.
Programs may need to be tuned for optimal performance. In particular, the use of the runtime option RPTSTG(ON), while useful when tuning, is much more costly now to leave on in a production program.
Recompiling all your PL/I source is recommended. If this isn't done, you need to carefully select the compiler options for compiling Enterprise PL/I code that will be mixed with older PL/I objects. You will also need to divide your source into partitions according to how they use FILEs, CONTROLLED variables and conditions. For more information, see "Object and load module considerations" in topic 4.9.1. |
If none of these apply, your best bet would be to open a PMR with IBM. |
|
| Back to top |
|
|
CICS fan
New User
Joined: 03 Apr 2008
Posts: 82
Location: United States
|
|
|
|
| We had to open a ticket with IBM. Thanks Robert. |
|
| Back to top |
|
|
CICS fan
New User
Joined: 03 Apr 2008
Posts: 82
Location: United States
|
|
|
|
the process is running in about 1/4th the time when compiled with PL/1 version 2.3
but on version 3.9 its taking more than 2 hours!
anyone here faced similar problem? |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hello,
Does this process still use about the same amount of cpu as previously?, the same amount of i/o? |
|
| Back to top |
|
|
CICS fan
New User
Joined: 03 Apr 2008
Posts: 82
Location: United States
|
|
|
|
| the cpu time is also equally slower like the elapsed time |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hello,
| Quote: |
| the cpu time is also equally slower like the elapsed time |
Suggest you update your PMR to include this info. . .
The increased cpu usage is very likely the reason the elapsed time is longer. |
|
| Back to top |
|
|
CICS fan
New User
Joined: 03 Apr 2008
Posts: 82
Location: United States
|
|
|
|
| ok, thanks. Will do that. |
|
| Back to top |
|
|
|
|
