The granularity depended on the model.
-- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3 ________________________________________ From: IBM Mainframe Discussion List <[email protected]> on behalf of Paul Gilmartin <[email protected]> Sent: Friday, February 23, 2018 3:13 PM To: [email protected] Subject: Timer Unis (was: ... time change ...) On Fri, 23 Feb 2018 14:54:18 -0500, Tony Harminc wrote: > >Timer Units are not TOD clock units. Timer units are approximately >26.04167 microseconds. They come from the long-gone S/360 Interval >Timer, which was the fullword at location 80 (x'50'). This was defined >so that bit position 23 is decremented every 1/300 second, which >conveniently allowed an implementation that decremented either bits 21 >and 22 every 1/50 second, or bits 21 and 23 every 1/60 second, thus >being able to run on a 50 or 60 Hz power line. Of course only the >smallest 360 models actually used the power frequency for timing, but >the definition lives on. > How accurate was that power frequency? ISTR that a few decades ago my electric clock would wander over long term, always remaining within NIST time ±15 seconds, so in long term it was better than crystal accuracy. I suppose that periodically the power company referred to USNO to make a correction. ("It depends.") Nowadays, it's much better. Was the granularity actually 1/60 or 1/50 second, or did it actually get finer granularity by using a PLL with line power as a reference? I suppose, "It depends." -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
