On 6 October 2013 15:10, [email protected] <[email protected]> wrote:
> I would load BPX1GHN on the first invocation and call BPX1GHN. Subsequent 
> invocations would bypass the LOAD and issue the CALL directly without issuing 
> another LOAD for BPX1GHN (see code sniper below).

Whatever your performance problem is, this isn't it. But even this
minuscule overhead can be eliminated, because BPX1GHN, like all the
UNIX Callable Services, is a tiny module containing five or so
instructions that load an address off the CVT and branch to it. What's
at that address is little more than a PC instruction. This is
documented in the Callable Services book; search for the word
"offsets".

> The issue I experienced, was that every call to BPX1GHN seem to take a long 
> time to perform. We are talking about 30 - 60 seconds per invocation of 
> BPX1GHN.

It is virtually impossible that this has anything to do with loading
or even execution of BPX1GHN. It is surely a network issue of some
sort. Can you run an IP trace to see what the timing of the outgoing
and incoming packets looks like on the wire?

Tony H.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to