Vector Packed Decimal instructions experiences
Just wondering if any shops have taken advantage of these new instructions on the Z14 and Z15 and if so, how is your real world experience? https://ibmsystemsmag.com/IBM-Z/01/2018/vector-facility-z14 Looking at IBM's benchmarks they look pretty good and could have a postive effect on reducing your 4HRA. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: Vector Packed Decimal instructions experiences
I believe the Enterprise COBOL 6 compiler generates code that takes advantage, assuming ARCH(12) is specified. If you are *not* specifying ARCH(nn) where 'nn' is the architecture level of your oldest machine (don't forget DR!) then you are leaving MSU's on the table. The ARCH(nn) level is two less than the "marketing number": a z14 is ARCH(12), and so forth. Charles -Original Message- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Laurence Chiu Sent: Thursday, April 23, 2020 7:01 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Vector Packed Decimal instructions experiences Just wondering if any shops have taken advantage of these new instructions on the Z14 and Z15 and if so, how is your real world experience? https://ibmsystemsmag.com/IBM-Z/01/2018/vector-facility-z14 Looking at IBM's benchmarks they look pretty good and could have a postive effect on reducing your 4HRA. -- 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
Re: Vector Packed Decimal instructions experiences
You are talking about packed unsigned math a that has been around in IBM CPCS since forever Sent from my iPhone > On Apr 23, 2020, at 21:09, Charles Mills wrote: > > I believe the Enterprise COBOL 6 compiler generates code that takes > advantage, assuming ARCH(12) is specified. > > If you are *not* specifying ARCH(nn) where 'nn' is the architecture level of > your oldest machine (don't forget DR!) then you are leaving MSU's on the > table. The ARCH(nn) level is two less than the "marketing number": a z14 is > ARCH(12), and so forth. > > Charles > > > -Original Message- > From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On > Behalf Of Laurence Chiu > Sent: Thursday, April 23, 2020 7:01 PM > To: IBM-MAIN@LISTSERV.UA.EDU > Subject: Vector Packed Decimal instructions experiences > > Just wondering if any shops have taken advantage of these new instructions > on the Z14 and Z15 and if so, how is your real world experience? > > https://ibmsystemsmag.com/IBM-Z/01/2018/vector-facility-z14 > > Looking at IBM's benchmarks they look pretty good and could have a postive > effect on reducing your 4HRA. > > -- > 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
Re: Vector Packed Decimal instructions experiences
In article you wrote: > Just wondering if any shops have taken advantage of these new instructions > on the Z14 and Z15 and if so, how is your real world experience? > https://ibmsystemsmag.com/IBM-Z/01/2018/vector-facility-z14 > Looking at IBM's benchmarks they look pretty good and could have a postive > effect on reducing your 4HRA. The z/Linux C compiler uses them. I was debugging a problem and was surprised to see the vector instructions used. In that case, it wasn't really needed as it was dealing with a single element, but perhaps the compiler generated the code to use the vector register rather than saving/ reloading a general register. I can't recall if I've seen them used with the z/OS C compiler, but it wouldn't surprise me. -- Don Poitras - SAS Development - SAS Institute Inc. - SAS Campus Drive sas...@sas.com (919) 531-5637Cary, NC 27513 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: Vector Packed Decimal instructions experiences
Packed decimal has been around since the beginning of time. Having dedicated instructions like the vector packed decimal ones only arrived with the Z14 and later mainframes,l And the point of the article I quoted was, with ARCH(12) and a Z14 Cobol code was timed as running up to 2x as fast as ARCH(11). So as has been said, if you have a Z14 or later and are not recompiling your code, you are potentially leaving MSU's on the table. On Fri, Apr 24, 2020 at 2:24 PM Steve Beaver wrote: > You are talking about packed unsigned math a that has been around in IBM > CPCS since forever > > Sent from my iPhone > > > On Apr 23, 2020, at 21:09, Charles Mills wrote: > > > > I believe the Enterprise COBOL 6 compiler generates code that takes > advantage, assuming ARCH(12) is specified. > > > > If you are *not* specifying ARCH(nn) where 'nn' is the architecture > level of your oldest machine (don't forget DR!) then you are leaving MSU's > on the table. The ARCH(nn) level is two less than the "marketing number": a > z14 is ARCH(12), and so forth. > > > > Charles > > > > > > -Original Message- > > From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] > On Behalf Of Laurence Chiu > > Sent: Thursday, April 23, 2020 7:01 PM > > To: IBM-MAIN@LISTSERV.UA.EDU > > Subject: Vector Packed Decimal instructions experiences > > > > Just wondering if any shops have taken advantage of these new > instructions > > on the Z14 and Z15 and if so, how is your real world experience? > > > > https://ibmsystemsmag.com/IBM-Z/01/2018/vector-facility-z14 > > > > Looking at IBM's benchmarks they look pretty good and could have a > postive > > effect on reducing your 4HRA. > > > > -- > > 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 > -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: Vector Packed Decimal instructions experiences
+1 Charles -Original Message- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Laurence Chiu Sent: Friday, April 24, 2020 3:09 PM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: Vector Packed Decimal instructions experiences Packed decimal has been around since the beginning of time. Having dedicated instructions like the vector packed decimal ones only arrived with the Z14 and later mainframes,l And the point of the article I quoted was, with ARCH(12) and a Z14 Cobol code was timed as running up to 2x as fast as ARCH(11). So as has been said, if you have a Z14 or later and are not recompiling your code, you are potentially leaving MSU's on the table. On Fri, Apr 24, 2020 at 2:24 PM Steve Beaver wrote: > You are talking about packed unsigned math a that has been around in IBM > CPCS since forever > > Sent from my iPhone > > > On Apr 23, 2020, at 21:09, Charles Mills wrote: > > > > I believe the Enterprise COBOL 6 compiler generates code that takes > advantage, assuming ARCH(12) is specified. > > > > If you are *not* specifying ARCH(nn) where 'nn' is the architecture > level of your oldest machine (don't forget DR!) then you are leaving MSU's > on the table. The ARCH(nn) level is two less than the "marketing number": a > z14 is ARCH(12), and so forth. > > > > Charles > > > > > > -Original Message- > > From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] > On Behalf Of Laurence Chiu > > Sent: Thursday, April 23, 2020 7:01 PM > > To: IBM-MAIN@LISTSERV.UA.EDU > > Subject: Vector Packed Decimal instructions experiences > > > > Just wondering if any shops have taken advantage of these new > instructions > > on the Z14 and Z15 and if so, how is your real world experience? > > > > https://ibmsystemsmag.com/IBM-Z/01/2018/vector-facility-z14 > > > > Looking at IBM's benchmarks they look pretty good and could have a > postive > > effect on reducing your 4HRA. > > > > -- > > 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 > -- 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
