Here is something interesting I found

http://www.gpsinformation.org/dale/nmea.htm#sirf

Look at NMEA sentence $PSRF104.

It appears there is a way to shorten the initial lock time by providing the SiRF chip with

- current location and altitude
- current time
- initialization mode:

1 = warm start, ephemeris valid
2 = clear ephemeris, warm start (First Fix)
3 = initialize with data, clear ephemeris
4 = cold start, clear all data
8 = cold start, set factory defaults

I am guessing it is gpsdriver who issues these sentences to the chip. Improving GPS performance may be as easy for Nokia as setting the right values in this sentence.

BTW, I noticed that the internal clock in my N810 is fairly inaccurate -- it loses 3-5 seconds per day. I wonder if this is a factor in GPS performance. Warm start requires accurate time, though I am not sure how accurate.

Interesting post.

I wouldn't think 3-5 seconds/day would make much of a difference - by the time you hit 'real' amounts of time, your ephemeris would be invalid.

It may be the gpsdriver, but the gpsdriver may just power the device on and off. gpsd is capable of sending NMEA sentences as well, and the maemo 2008 gpsd seems to be a maemo fork (based off an ancient gpsd, from the version number: 1.0-23)

I can't check the link as it now looks like a squatted domain.

as far as time goes, gpsd could reset the time if started with the right flag and ntpdate was configured properly. Otherwise, it is an _internet_ tablet - why not just include ntpdate running under anachron by default?

Can't be a reason for the long TTF on the N810, unless the gpsdriver already got updated between the N810 release and N800 OS2008 beta - on the N800 with LD3-W, the TTF is reasonably fast.