Back in June 2018 there was discussion on this list that indicated that the forward progress on a subcapacity processor is suspended sufficiently often for a small time interval (4 µsec?) to degrade the average SU capacity of the processor to the licensed value. It would seem logical to assume such suspension would have to occur on clock cycle boundaries and be for one or more clock cycles.
That would mean that while the processor is making progress, it would be running at full clock speed, but there would be some clock cycles where no progress occurred. From the users perspective and program code execution speed perspective it would appear as if the effective clock speed were reduced even though the physical clock speed was unchanged. This was the same design concept used to produce a sub-capacity IBM 407 Accounting machine back in the 1950's: the IBM 407 ran at 150 cards/minute, but a degraded and presumably cheaper-to-lease model was available that paused every third machine cycle, reducing the effective speed to 100 cards/minute. Joel C. Ewing On 3/3/19 10:07 AM, Christopher Y. Blaicher wrote: > ZIIP and ZAAP processors always run at full speed, even when running on a > sub-capacity box. > One thing, among many, I don't know is how IBM implements sub-capacity. Slow > the clock speed? Skip cycles? > > Chris Blaicher > Technical Architect > Syncsort, Inc. > > > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On > Behalf Of Parwez > Sent: Sunday, March 3, 2019 6:05 AM > To: IBM-MAIN@LISTSERV.UA.EDU > Subject: Re: CPU time and zIIP > > One regular misconception about (cycle time), irrespective of the type of > processor the 'speed/cycle time' of ALL the processors is the SAME. CPs with > CAPACITY setting of 7xx (Zxx in case of the 'BC' class system) are FULL > capacity. Others are sub-capacity. So if the workload system hasn't enough > capacity, then it might not run as well as one with abundance. So from a > cycle point (speed?), zIIPs are not faster. > > Parwez Hamid > > ... -- Joel C. Ewing ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
