Re: most efficient method to determine "used cpu time" within a particular subtask
Frankly, one could always make an argument for using a supported API in all but the most extreme of situations. Charles -Original Message- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@bama.ua.edu] On Behalf Of Walt Farrell Sent: Wednesday, July 13, 2011 10:26 AM To: IBM-MAIN@bama.ua.edu Subject: Re: most efficient method to determine "used cpu time" within a particular subtask >OTOH if you have a more normal sort of some IO/some computing/some >system services type of program and would just like to know roughly how >much CPU time you have consumed up to this point, then TCBTIME should >be totally satisfactory. But if you have that kind of program, and a rough answer would suffice, do you really need to figure out the most efficient way of getting it? (Conversely, if you do want the most efficient way, doesn't that (perhaps) imply that you don't have that kind of program?) -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: most efficient method to determine "used cpu time" within a particular subtask
On Wed, 13 Jul 2011 09:26:27 -0500 Walt Farrell wrote: > But if you have that kind of program, and a rough answer would > suffice, do you really need to figure out the most efficient way of > getting it? (Conversely, if you do want the most efficient way, > doesn't that (perhaps) imply that you don't have that kind of > program?) Personally, I would consider the pursuit of "most efficient" a most worthy aim - always. Mindlessly chasing "exact" might be another matter. Shane ... -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: most efficient method to determine "used cpu time" within a particular subtask
On Wed, 13 Jul 2011 09:29:51 -0400, Charles Mills wrote: >> if you look at system control blocks (such as TCBTTIME) you will not pick >up the time since the last dispatch. > >Which may or may not be satisfactory. > >If you are computing pi to a million digits on an otherwise idle machine and >would like to know how much CPU time each pass through the main loop >consumes, then TCBTIME is totally unsatisfactory. > >OTOH if you have a more normal sort of some IO/some computing/some system >services type of program and would just like to know roughly how much CPU >time you have consumed up to this point, then TCBTIME should be totally >satisfactory. But if you have that kind of program, and a rough answer would suffice, do you really need to figure out the most efficient way of getting it? (Conversely, if you do want the most efficient way, doesn't that (perhaps) imply that you don't have that kind of program?) -- Walt -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: most efficient method to determine "used cpu time" within a particular subtask
> if you look at system control blocks (such as TCBTTIME) you will not pick up the time since the last dispatch. Which may or may not be satisfactory. If you are computing pi to a million digits on an otherwise idle machine and would like to know how much CPU time each pass through the main loop consumes, then TCBTIME is totally unsatisfactory. OTOH if you have a more normal sort of some IO/some computing/some system services type of program and would just like to know roughly how much CPU time you have consumed up to this point, then TCBTIME should be totally satisfactory. IMHO Charles -Original Message- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@bama.ua.edu] On Behalf Of Peter Relson Sent: Wednesday, July 13, 2011 8:56 AM To: IBM-MAIN@bama.ua.edu Subject: Re: most efficient method to determine "used cpu time" within a particular subtask If by "within a particular subtask" you mean "not my task, but a subtask of it", then you're mostly on your own. There are no programming interfaces provided to accomplish that. And no control block fields that contain time for a task are programming interfaces. If you're interested in time for "my task", and if you happen to be running on a machine with the ECTG instruction available, then TIMEUSED ECTG=YES,STORADR=x is far and away the best. The output is not in "MIC", it will be in TOD clock units. Alternately, TIMEUSED ECTG=COND,STORADR=x,LINKAGE=SYSTEM can be used. It will get the fast path if ECTG is available, and a slower PC-entered path if ECTG is not available. As Binyamin correctly pointed out, if you look at system control blocks (such as TCBTTIME) you will not pick up the time since the last dispatch. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: most efficient method to determine "used cpu time" within a particular subtask
If by "within a particular subtask" you mean "not my task, but a subtask of it", then you're mostly on your own. There are no programming interfaces provided to accomplish that. And no control block fields that contain time for a task are programming interfaces. If you're interested in time for "my task", and if you happen to be running on a machine with the ECTG instruction available, then TIMEUSED ECTG=YES,STORADR=x is far and away the best. The output is not in "MIC", it will be in TOD clock units. Alternately, TIMEUSED ECTG=COND,STORADR=x,LINKAGE=SYSTEM can be used. It will get the fast path if ECTG is available, and a slower PC-entered path if ECTG is not available. As Binyamin correctly pointed out, if you look at system control blocks (such as TCBTTIME) you will not pick up the time since the last dispatch. Peter Relson z/OS Core Technology Design -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: most efficient method to determine "used cpu time" within a particular subtask
On Tue, 12 Jul 2011 07:48:39 +0200 "Dr. Stephen Fedtke" wrote: :>several sources provide "used cpu time". calculating the difference between :>two points of time allows to calculate the required cpu time for particular :>parts of the processing. :>question: which is the most efficient and reliable method to determine the :>"cpu time used" in order to apply the above given delta calculation? is :>there anything more efficient than call a service, such as :>TIMEUSED CPU=MIC,STORADR=WORK_UCPU, * :> LINKAGE=SYSTEM :>for example, reading out any control block fields. The problem with control block fields is that they are only updated when a decision is made to dispatch a different unit of work. Thus they will only show figures valid up to the last dispatch. -- Binyamin Dissen http://www.dissensoftware.com Director, Dissen Software, Bar & Grill - Israel Should you use the mailblocks package and expect a response from me, you should preauthorize the dissensoftware.com domain. I very rarely bother responding to challenge/response systems, especially those from irresponsible companies. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
most efficient method to determine "used cpu time" within a particular subtask
hi all, several sources provide "used cpu time". calculating the difference between two points of time allows to calculate the required cpu time for particular parts of the processing. question: which is the most efficient and reliable method to determine the "cpu time used" in order to apply the above given delta calculation? is there anything more efficient than call a service, such as TIMEUSED CPU=MIC,STORADR=WORK_UCPU, * LINKAGE=SYSTEM for example, reading out any control block fields. best stephen --- Dr. Stephen Fedtke Enterprise-IT-Security.com Seestrasse 3a CH-6300 Zug Switzerland Tel. ++41-(0)41-710-4005 www.enterprise-it-security.com ++NEWS++ SF-NoEvasion lets you avoid all 10 pitfalls when connecting z/OS to your SIEM ++NEWS++ -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html