Page 2 of 2

Re: Program Fetch Performance

PostPosted: Tue Jul 02, 2013 9:54 pm
by dick scherrer
You're welcome - though, i wasn't near as much help as nevilh :)

I see nothing amiss with my name. Dick is fine - Mr. was my dad 8-)

d

Re: Program Fetch Performance

PostPosted: Wed Jul 03, 2013 1:21 am
by steve-myers
Well, it was you that said
Is it possible that the pdse load only actually read dasd once and then used the previously read module?
That didn't lead anywhere, but with some prodding by nevilh, this ultimately led to the manual and the FETCHOPT discussion. I was aware of the option, but since I so rarely use PDSE, I had never studied the description in detail. I still don't really understand the PACK/NOPACK part of FETCHOPT.

This is the timing part of the program
*        TIMEUSED STORADR=STARTTOD,CPU=TOD,LINKAGE=SYSTEM   
         STCK  STARTTOD            SAVE THE START TIME     
         L     3,REPS              LOAD NUMER OF REPETITIONS
RUNLOAD  LOAD  EPLOC=PGM           LOAD THE LOAD MODULE     
         DELETE EPLOC=PGM          DELETE THE LOAD MODULE   
         BCT   3,RUNLOAD           TRYM TRY AGAIN           
*        TIMEUSED STORADR=STOPTOD,CPU=TOD,LINKAGE=SYSTEM   
         STCK  STOPTOD             SAVE THE END TIME

My original thought when I started my mini project was to study CPU, but I quickly decided that was silly and just concentrated on elapsed time. Contents Management is not supposed to retain the module when doing LOAD/DELETE as shown, though PDSE might have cached the module in some fashion. Non TSO OS/360 might have held onto the module, but that "feature" was removed in MVS.

As for the Mr. stuff. I am just trying to be formally courteous.

Re: Program Fetch Performance

PostPosted: Wed Jul 03, 2013 3:28 am
by dick scherrer
Appreciated - but not really necessary. I'm good with whatever works for you ;)

d