I think I disagree. You compile the program for ARCH(8). IBM guarantees that it will run on a z10 (do I have that right?). They do NOT guarantee that the program plus LE will behave on a z114 exactly as though it were running on a z10.
No matter what ARCH the program were compiled for, I would expect that LE running on a z114 might well exploit the actual hardware. I would be kind of unhappy if it did NOT. The vendor product either supports z114's or it does not. If they do not support z114 instructions, they should admit that they do not. > If LE really is doing this, why even have an ABO product To update ("optimize") the *compiled* object code. The OS-resident support/library modules (LE) are a different matter. They are already (I am guessing) at a current level. What is the z/OS release? I would expect LE to be built for the lowest level hardware that that release supported, but LE might be clever enough to dual-path, and I think that would be a good thing. Charles -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Mark Zelden Sent: Friday, May 3, 2019 3:35 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: COBOL 6.2 and ARCH(12) On Fri, 3 May 2019 15:57:34 -0400, Brian Chapman <bchapma...@gmail.com> wrote: >We have a vendor debugging product that is constantly causing 0C1 and 0C4 >abends since we have upgraded to COBOL 6.2. It also caused these abends >when we were at COBOL 4,2, but the abend rate has grown considerably after >the upgrade. > >The vendor has produced countless patches, but so far they have not >resolved the issues. We were notified today that they believe they >understand the issue. They are stating that even though our COBOL compiler >is set with ARCH(8) (to support our DRE machine), LE run-time is >recognizing that the program is COBOL 6.2, running on a z14, and >automatically switch the ARCH level to ARCH(12). They believe the run-time >execution is exploiting the new Vector Packed Decimal Facility and >producing erratic behavior. > >I searched through several presentations and IBM manuals for COBOL 6.2, and >everything I have found states that a recompile with ARCH(12) is required >to take advantage of the new facility. Is the vendor correct? > > I've never heard of that and I wouldn't expect IBM to ever do something like that, but heck, what do I know. ;-) LE shouldn't be trying to outsmart the person that compiled the code (IMHO). 1) Have you verified the options in a compile listing are as you expected? 2) Are you running ABO and could that be involved? Although I know nothing about configuration ABO (I have never "seen" or used it), even if you were I woudn't think you would have it configured to use z14 instructions. If LE really is doing this, why even have an ABO product. I certainly would open an SR with IBM LE support about it. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN