Re: CPU Percentage Flex, Lies, and Usage Measuring Tape

DAsmussen@aol.com wrote:
> 
> Hey Folks,
> 
> Now it's my turn to ask a couple of questions.  Since our upgrade to 3.7
> (production) and 3.2 (development system), jobs no longer seem to get the CPU
> that they deserve when running independently of the users.  Case in point,
> I'm running our BPCS shop order purge this weekend which removes data from
> approximately 23 files.  Prior to our upgrade from 3.1, this job (and
> rightfully so) consumed about 80% of available CPU at normal batch processing
> priority on our production 320.  This weekend, I'm running the purge on both
> our new production 530 (3.7) and our newly annointed development 320 (3.2).
>  Despite my upping both jobs to priority 10, setting purge for both to *NO,
> and setting both timeslices to 100,000, neither is consuming more than 18%
> CPU with nobody else on the systems.
> 
> Last week, we asked our Systems Engineering department for some help with our
> BPCS conversion program's performance under the same circumstances, and they
> did _something_ to make it run with more CPU percentage.  

Increased the jobs TIMESLICE value or RUNPTY, and maybe there were other
competing jobs running at the same time and therefore an increase may
have helped here ?

> I'm not sure that
> our manager actually asked, but we were not informed of what Systems
> Engineering actually DID to make this program perform better.  In addition, a
> program with a loop that was being tested on the (formerly development) 310
> took over the CPU, but had negligible impact when we moved to the new 320
> CPU.  Does anyone know of anything on the new OS versions that limits CPU
> percentages (none of the old limits were in place)?

Check system value QDYNPTYSCD on RISC system.
What is sysval QPFRADJ set to now vs. then ?
Is the 'expert cache' (Paging option *CALC, from F11 on WRKSYSSTS) set
for the pool you are running the batch job in ?
Changing you priority to 10 isn't going to do a thing for you if there
are no other jobs competing for the CPU between priorities 10 and 50 at
the same time.

Never mind the CPU usage of the job, how long is the job taking to
complete now vs. what it took before ?  Is it still quicker than it was?
With a faster CPU on the new machine a job that may have been CPU bound
could now become disk bound, and unable to eat as many of the new faster
CPU cycles as it it could the slower cycles on the old system.
Were there any changes made to the disk configurations ?


> 
> Secondly, I heard a NASTY rumor that clouds my former anticipation of IBM's
> August announcements.  In addition to the 12-way processor, I also hear that
> IBM is dropping user-based pricing (which helped the AS/400 get as big as it
> is) in favor of returning exclusively to tier-based pricing.  Anyone have
> information on this?
> 

No.  But as they never successfully implemented the usage based pricing
(ie. ENFORCED with software keys) then anything is possible.


-- 
... Neil Palmer                                      AS/400~~~~~      
... NxTrend Technology -Canada       ____________          ___  ~     
... Markham,  Ontario,  Canada       |OOOOOOOOOO| ________  o|__||=   
... Phone: (905) 731-9000 x238       |__________|_|______|_|______)   
... url:http://www.NxTrend.com        oo      oo   oo  oo   OOOo=o\   
... mailto:NPalmer@NxTrend.com     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* This is the Midrange System Mailing List!  To submit a new message,   *
* send your mail to "MIDRANGE-L@midrange.com".  To unsubscribe from     *
* this list send email to MAJORDOMO@midrange.com and specify            *
* 'unsubscribe MIDRANGE-L' in the body of your message.  Questions      *
* should be directed to the list owner / operator: david@midrange.com   *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *