Re: CallP Versus Program 'CALL'

... first of all: talking about performance of call operations, we are talking about the diffrence between nano seconds and micro seconds and you would have to make kilo of calls to make the diffrence measurable and even then, you would not notice any diffrence in a real world applicaton.
Going to some details: the time consuming part of a call operation is the activation of a programm (or SRVPGM), once activated, there would be no typical diffrence between calling an activated main (= programm call) or calling a bound procedure, or calling an internal subprocedure, or branching to a subroutine - it depends on other factors what would be fastest. The time, taht is needed to activate a component depends on the size of this component, how many static variables need to be allocated and initialized and how many dependent objects have to be activated. In many cases activation of a SRVPGM might take longer, than activation of a standalone programm. Bound procedure calls may have the advantage, that with one activation, multiple procedures are bound, so that one activation enables some thousands of fast call operations, but there are scenarios where simple dynamic calls (calling the same programm 1000 times without setting LR) beats the bound call of a procedure (calling a procedure in a big SRVPGM in ACTGRP(*NEW)).

The important benefit of procedures and SRVPGMs is the possibility to have a programm object, with multiple entry points, having multiple procedures, sharing a statefull global context.

D*B