Hi
I appreciate your answer, and understand your point
I have just seen the REXX (System REXX etc ? ) IRXINIT dump, and looked
into the SYS1.SHASSRC.
On 4/12/2012 4:49 PM, John Eells wrote:
Miklos Szigetvari wrote:
Hi
We try to modernize our code here, with relative instructions, long
displacements , immediate s etc etc
What about the control program ?
Just got some REXX IRXINIT dumps, and seems to me the code is not very
modern.
<snip>
If you look at the past several releases' worth of LSPRs, you will see
that we *do* work on the performance of the BCP and other z/OS
components. But we prioritize what we do to maximize the overall
system performance return you (and we) get for our efforts. There
doesn't seem to be much value in "modernization" for its own sake, or
in expending effort on code paths that are relatively rarely traveled
and that work fine today. (We do have to test this stuff too, after
all.) Also, when optimization requires data area restructuring,
particularly for older data areas, it can be disruptive in terms of
compatibility.
Other than what you can see in the LSPR and in capacity planning
tools, this is an "invisible" effort when we do it right, but there
can be real benefits. If you can move down the n-way curve as a
result of improvements we make, for example, you can gain capacity
both from the performance improvements themselves and from a reduction
of n-way overheads.
This is not to say that you will not have any "hot spots" on your
systems in z/OS code which, if optimized, could help one or more of
your particular workloads, just that they haven't risen to a level
where we've seen them as important enough at an overall system level
to work on them. But "backwater" code that lives far away from any
frequently-traveled mainstream code path is an unlikely optimization
target.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
