Especially on small circles like 200m track. Look on the image! Actually, it is the worst image I ever seen on Strava! Compare my new 5 plus (on the top) and iPhone (bottom). Even cheapest Android costs 10x less than Fenix has much better GPS performance. Even my old Vivocative HR had MUCH better tracks. Now I shame to publish my runs from my 700 euro device, they look just terrible. I can't understand how Garmin was able to make a product with SO bad performance of its MAJOR feature.
до покупки F5+ пользовался 735,проблем с GPS не испытывал, решил взять F5+ и пожалел, через раз получал трек (вид занятия бег) как на картинке обведённой жёлтым, дико бесило, но нашёл причину плохих треков…
The reason why I publish this is my hope that eventually Garmin will hear us and finally will improve GPS performance. I have read that Polar made great FW GPS-related improvements so now they works nearly…
Are you in Russia? What latitude are you at? Have you tried GPS + Glonass instead of GPS + Galileo?
I ask because Glonass is supposed to have an advantage at high latitudes because the satellite orbits…
So Ive just finished the Blackmores Sydney Half Marathon today and logged a 22.39km run over what should have been 21.098 whilst using GPS+Galileo. Something's clearly gone wrong. Looking at the plots, we see an excellent visual demonstration on the impact of tall structures on perceived GPS coords. You can see the reflectance of the buildings in the first screen cap as we were moving through the CBD and a comparatively accurate plot in the second during the park run section.
I've tried configuring GPS to lock to roads and this gives me an accurate plot on the watches map, but still results in a skewed pace and errant GPS data when uploading to Strava. It's a bit of a shame Garmin can't seem to combine the accelerometer and compass data to validate the delta from the last coord and discard the obviously anomalous readings such as those seen on the first screen cap.
So finally I've installed 2.51 beta GPS update. Tracks looks better now in some extent; meaning not so ugly as used to be but still not so realistic as a track from smartphone or Polar. On other hand, on the 200m intervals it seems F5+ correctly counts laps. I also compared a plot after update from the latest parkrun with the plot from M400. Definitely, F5+ looks better now (note M400 out of the track places marked by red arrows). So 2.51 is not HUGE step forward but considerable; now my F5+ is not on sale anymore 
Thank you Alan. Do you mean this part? It looks good for me, doesn't for you? Two laps however probably is not so representative. Note that my track has proper "shape" overall. The problem comes with counts - shape becomes a ball as wrong laps accumulate.
до покупки F5+ пользовался 735,проблем с GPS не испытывал, решил взять F5+ и пожалел, через раз получал трек (вид занятия бег) как на картинке обведённой жёлтым, дико бесило, но нашёл причину плохих треков и стал получать как на картинке слева, вся причина - это в каком направлении "смотрят" часы, так 
или так 
короче, если часы все время смотрят вверх проблем ни с треком ни с измеряемым темпом нет, как только в пол и в стены зданий - записанная тренировка на смарку.
For those who can't speak Russian, 9675982 mentioned that he could only get accurate tracking when he wore the watch facing upwards such as he is in the second image 
