GPS Computer Speed Accuracy

[Alien8]

But as aforementioned, range-rate between two points on an undulating surface will differ between a line drawn in 3D space between the same two points. Even assuming both devices were 100% accurate the results will still differ, because they're measuring different things.

I don't get your point @Drago, GPS measured range-rate is an instantaneous measurement of the frequency shift in the received signals - differences in position, straight-line or otherwise, has no relevance in this respect and has no impact on GPS measured velocity.

Anyway, as has already been pointed-out elsewhere, there are a myriad of factors that can influence the velocity as displayed to the user and hence any perceived difference, wheel-revolution-counting vs GPS, is probably not worth worrying about.

 

[Ajax Bay]

or if you are aiming to catch a train

The last time I felt I had to pay attention to speed per se, was when I had 9 hours to complete the last 100 miles of a 1014km audax (in July), starting the last half day at 4am. The key readout I used was not 'real time' speed but average speed and its progress/maintenance above the 18kph threshold (which would get me to the finish in time). I made it with half an hour to spare - 74 1/2 hours in total.
On my Garmin I have a '5 window' display, with heart rate as the primary, and also showing speed, average speed, distance and cadence.
On my wired Cateye Velo 2 it has speed as primary with arrows showing whether that speed is above or below average (this primary display is standard and not changeable) and normally time of day as the subordinate.