Yes, with STFL.
-- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3 ________________________________________ From: IBM Mainframe Discussion List [IBM-MAIN@LISTSERV.UA.EDU] on behalf of Ituriel do Neto [000003427ec2837d-dmarc-requ...@listserv.ua.edu] Sent: Wednesday, March 15, 2023 9:30 AM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: CS/CDS instruction Can we detect if a specific feature is available in the current hardware? Best Regards Ituriel do Nascimento Neto z/OS System Programmer Em sábado, 11 de março de 2023 às 14:05:12 BRT, Paul Gilmartin <0000042bfe9c879d-dmarc-requ...@listserv.ua.edu> escreveu: On Sat, 11 Mar 2023 00:03:06 -0800, Leonard D Woren wrote: >If some particular instruction set feature is installed, the >definition of ASI/AGSI is enhanced to serialize the update, making it >a simpler solution than a CDS loop or PLO. > >In some performance testing a while back on a z14 or z15 which I think >had the above serialization feature, the execution times for a very >large number of executions of L / AHI / ST were very close to the same >count of ASI. If I recall, the ASI was a few percent slower, I guess >because of the serialization. I.e., unless you're doing abnormal >tests as I did, you won't notice the difference. > >From the PoOps (excerpted): The storage-operand update reference for the follow- ing instructions appears to be an interlocked-update reference as observed by other CPUs and channel programs. • TEST AND SET • COMPARE AND SWAP (of course) • AND (NI and NIY), when the interlocked-access facility 2 is installed • OR (OI and OIY), when the interlocked-access facility 2 is installed (at last!) • ADD IMMEDIATE (ASI and AGSI), when the interlocked-access facility 1 is installed and the first operand is aligned on an integral boundary corresponding to its size The feature-dependent instructions are treacherous. Programmers must avoid them in multi-tasking code intended to be portable. I consider it bad hardware design to introduce feature-dependent instructions. They should have remained invalid operations on models lacking the interlock facility. I believe NI and OI are primeval: they antedate multiprocessors and became unsafe only at the advent of multiprocessors. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
