On Sun, 2020-03-08 at 17:59 +0000, Chris Olson via CentOS wrote: > A few years ago, one of our interns was curious about system > time keeping features in computer systems. This intern was > also the proud owner of an inexpensive Radio-Controlled Clock. > The intern wondered why computer motherboards were not just > equipped with a chip like the ones in the RCC so that their > system time would always be correct. > > > This large response was a bit of a surprise and included a > bunch of time related horror stories. It became clear why > using an RCC chip on motherboards would NOT be a good idea. > GPS network time servers seemed to be a preferred choice. > The problem with radio time signals is that they just aren't accurate enough. Your bedroom clock needs to be correct to within a minute or so, but they are generally correct to about +/- 5 seconds. That's just not good enough for system times. There's also a massive problem with signal strength in the UK - the (singular) time transmitter is in the middle of the country in Cumbria and in the south it's virtually impossible getting a signal any further than about 2 feet from a window - not a hope of getting anything in an office building! GPS times also have problems. They are very accurately wrong! The atomic clocks on the satellites haven't been updated since they were launched, so no leap seconds. There are corrections that can be applied once the time has been received but it depends on a knowledge of leap seconds - I think they are currently about 18 seconds out. But they are accurate to about 10-100ns. You also need a decent antenna to get the high accuracy, which again means that you need to be near a window to see the satellites. Generally much easier to use NTP! P.