The performance benefit of NOCLOSE is startup time, because it doesn't have to open/close SYSEXEC each time. The best performance is achieved if you invoke ISPF *from* an exec, so SYSEXEC is always open -- just like the other ISPF DDs.
There's no improvement to execution performance of the execs. Its just the load time. But it makes a noticeable difference, especially if your ISPF system has lots of custom menus leading to REXX execs. -----Original Message----- From: IBM Mainframe Discussion List <[email protected]> On Behalf Of Paul Gilmartin Sent: Friday, July 12, 2024 1:50 PM To: [email protected] Subject: Re: IRXJCL with sequential SYSEXEC On Fri, 12 Jul 2024 18:31:19 +0000, Schmitt, Michael wrote: >It is because on MVS, REXX execs can be mixed in with CLISTs in the SYSPROC >DD. In that case, it needs the comment to know that it is REXX. (Assuming >EXECUTIL SEARCHDD(YES) is in effect, which it is by default I think.) > >This is one reason why putting REXX execs in SYSEXEC improves performance. > Something I noticed but never reported is that TABs (x'05') may be used for indention is execs loaded from SYSPrOC. In SYSEXEC they are syntax errors. >The other is that you can get a TREMENDOUS boost in performance by setting >EXECUTIL EXECDD(NOCLOSE). It is like 20 times faster, the last time I measured >it. > Is that startup overhead, or does it apply throughout a long-runnibng compute-intensive EXEC? >** BUT THERE ARE CONSEQUENCES. ** > >Don't do this unless you are prepared to deal with them. > >I've had our team running with NOCLOSE for decades, so they're not >insurmountable. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
