Time is often off by about 3 minutes after just few days without GPS

That's interesting. A few seconds isn't bad. Casio watches typically also lose several seconds every month. I wouldn't notice that. 3 minutes was for me after just 2 weeks but it seems it is quite abrupt. I should track this now daily to see if this is constant or sudden.

I also wonder if this issue is related to slow performance of watchfaces on connect including Garmin own store watchfaces. It seems to me the OS software is struggling with performance and is in need of more optimization as previous models. Reason for now that I'm using the default watchfaces only.

gaijin said:

The more the watch connects with Wi-Fi, phone, or GPS, the more accurate the displayed time will be.

In my experience, on multiple Instincts, there's only one way to get an accurate time sync and that is via the GPS. Syncing via the phone doesn't adjust the time in the same way as the GPS, it only adjusts it to make it more aligned within the ballpark but not down to a second. Which is frustrating because I don't know any other watch brand which would offer a watch with a bluetooth connectivity and not utilise it to sync the time perfectly - even once per day, FFS. 

I tested this now for 2 weeks without any GPS or phone sync (or WiFi) and time is still accurate within 1 second which is impressive.

So I think something caused the time to be off by 3 minutes (especially since it happened 3 times now). Most likely it was the slow long update. The other option could be turning it off at night which I don't do now anymore but that seems more unlikely.

The time gets practically always off a few minutes after every firmware update. The watch seems to keep the time from the moment when the update starts, and stops counting while updating, which usually takes a few minutes.

That's one of reasons why I perform a hard reset after any update, and sync the watch (the syncing with the GC app syncs the time too), though there are a few more reasons for doing it, too. Syncing the time over GPS is even a better option, though it happens automatically at the next outdoor activity.

gaijin said:

Regular connection with phone, wi-fi and GPS keeps the variance well within these limits - usually within a second.

Having read the whole thread I am happy that at least a few people spent some time with the (in)accuracy of the internal clock of the watch. Although any outdoors activity with GPS data means a quasi continuous option for GPS time sync, my finding was that maybe this time sync does not always happen during activities. 

I identified a time slip of 7-9 seconds during 284 minutes. I am not sure whther it was 7 or 9, because I do not know whether the competition organizer had an accurate clock or youtube video length is accurate. But i say it was 7 or 9, and not zero, because I was the only one from a groupe of fellow runners who experienced a difference of more than 2 sec  with his Garmin watch (Enduro 2).  

forums.garmin.com/.../1984073

Do you have any bet how timestamps of activity could have a time slip if one set GPS ON? I mean there was a couple of minutes when GPS data were not available, but it would be so strange that it resulted in a time slip of 7-9 seconds during 4.75 hours.

They most likely sync time only once, at the start of an activity. After that it relies on the watch regulating time correctly. Same as the elevation - it's not synced continuously, it relies on the sensor tracking the ascent and descent (which is also not very accurate, you can enable topo maps based elevation for an activity to see that).

Issue with timekeeping is quite simple, these watches don't have any internal time regulator. They run on CPU cycles and anything can knock them out of whack. This maybe wouldn't be a problem, if they kept syncing time perfectly to a second every time there's a chance to do it (be it live GPS or sync with the phone). But apparently in Garmin's view their watches don't have to perform this core functionality correctly.

Thanks for sharing your thoughts with me. You did not make me happy, 

You mentioned the total ascent/descent data ,and to be honest I think Garmin at least tried to do their best as regards the altitude/ascent/descent accuracy.

25 (?) years ago (I am not sure which  Garmin device was the first  with barometer, eg. the popular 60cs was released in 2003, but Etrex Summit already in 2000 ) when they first added barometer to a device to increase the altitude accuracy and so that of total ascent they shortly figured out that the inaccurate GPS elevation helps in some situation to the generally more accurate barometer based altitude.

Precisely when the weather induced altitude drift seems to appear then the avg data of GPS elevation starts to behave as a glue or more like a rubber band to avoid that the altimeter error would go over a threshold due to environmental pressure change.

It is an early (2001) Garmin patent about barometric altitude: https://patentimages.storage.googleapis.com/b5/fa/3a/aebf34bdc779bc/US6735542.pdf

And it is from 2006 what I shortly described above: patentimages.storage.googleapis.com/.../US7429948.pdf

And a bit more than a decade later Garmin added DEM elevation to the tools to be used when altitude is calculated.

And we speak about doubt of the time accuracy of Garmin watches in 2025. 

These two things putting next to esch other is so surreal.

I only have seen time being off after a FW update (seen that about 3 times and quick time sync after fixed it for me). I checked after about 2 weeks and was still pretty accurate.

I personally haven't encountered any time sync issues after recording more than a dozen navigation tracks (but all 1 activity with all same settings). I could imagine there could be bug or crashes causing time to be off for certain activities and settings. Best to collect the fit file for debugging and/or file a bug.

Luc said:

Best to collect the fit file for debugging

If you had read my post in the other thread I referred to you would have  known that I made some deep digging in the fit file also like comparing all the neighbouring timestamps looking for  eg. identical location data. And  only after I had run out of ideas I started to post about it in Garmin forums.

I found nothing and even when I had no gps signal during my marathon (long tunnel) and there were no location data I could not find any neighbouring timestamps where all the other  data (speed, cadence, power, hr) were the same. I mean I found zero instances of duplicated data in two timestamps following each other.

And I found no missing timestamps, in each second there were a record in the .fit file. Etc.