2 ms·
To be anal: technically GMT is closer to UT1. UTC is never "known" until after the time you are referring to as it is the reading from an atomic clock with some
by dspillett 16y ago
To be anal: technically GMT is closer to UT1. UTC is never "known" until after the time you are referring to as it is the reading from an atomic clock with some leap-seconds. The exact adjustment is not decided (though can generally be predicted) until IERS publishes DU1 for the time.
Though if you are not needing long period times more accurate than +/- 0.9s you can consider GMT, UT1 and UTC to be equivalent, and for sub-second measures you need to reference a local timer. This is essentially how NTPd works too - it keeps a local timer with reference to hardware events like timer interrupts plus some skew values and this is what your OS ends up using to track small amounts of time, and readings from external clocks are used to tweak the skew values so that this local timer doesn't fall too far out of sync. The skew relationship between UT1 and UTC is called DUT1.
See http://en.wikipedia.org/wiki/Coordinated_Universal_Time http://en.wikipedia.org/wiki/Coordinated_Universal_Time for more than you care to know about UTC (and, if you do care that much or simply have some serious procrastinating to work on, other references are listed too for further reading).
On London time not being GMT for half of the year, this can be a hassle. We are UK based as are our current clients for the most part. One of the larger ones (a major bank, but for contractual reasons I should probably refrain from mentioning their name) have their email and calendaring system set to GMT, which means that any time they invite us to a meeting or teleconference in this half of the year we have to adjust the time (as our mail+calendar arrangement are correctly set for UK time with daylight saving accounted for, as are the systems of our other clients) or turn up an hour late. Most irritating.