Mark Zelden wrote:
On Mon, 5 Mar 2007 07:41:15 -0500, John Eells <[EMAIL PROTECTED]> wrote:
This makes perfect sense to me. Why worry about which cases
matter when you can always use an action that's not harmful and
covers all of them?
Because some people actually care about speed of execution and CPU time.
z9 cycles don't come cheap. Those of us trying to blead every mip out
our boxes in order to save our beloved mainframe (and our jobs) have
to care about these kinds of things.
PDSFAST wouldn't exist if IEBCOPY was quick and efficient. COPYMOD
is especially slow.
Example for just 1 library:
COPY of SYS1.LINKLIB to new library (same blksize):
CPU 0MIN 01.00SEC SRB 0MIN 00.93SEC (35 sec wall clock)
COPYMOD of SYS1.LINKLIB to new library (same blksize):
CPU 0MIN 03.36SEC SRB 0MIN 00.94SEC (2 min 28 sec wall clock)
<snip>
I had not considered that. It would be interesting to know, if
one could, how much overhead would be saved *after* a COPYMOD,
though. The COPYMOD itself might be expensive, but the potential
downstream reduction in CCW chaining (which also has a CPU cost)
and fetch time (which has a CPU cost when Program Fetch's PCIs
aren't timely and a CCW chain must be rebuilt) might prove worth it.
Without knowing the benefit, it's hard to say whether the cost is
justified.
--
John Eells
z/OS Technical Marketing
IBM Poughkeepsie
[EMAIL PROTECTED]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
