I'm struggling to find a way to deal with the Alphas that GW52s give.

When running at 5Hz, they have 5 times as many errors as any normal 1s GPS file. Any changes I make to accommodate the extra errors that come with GW52s causes spikes in GT-31s to be ignored and false results with those.

Any suggestions?

The GW52 file certainly isn't as smooth as the GT31's but I'm not sure they're errors, Andrew thinks the bumps are real, I certainly get a smoother result when sailing on dead flat water.
Can you use a frequency filter, 5hz bumps allowed, anything else rejected?

Well the latest incarnation of ka72 seems to be doing better than I thought. I can't find any bad examples of GW52 data now. (I invite everyone else to send me any Alpha files that don't calculate right.)

On the other hand, the 1hr calculation is frustratingly slow. I need to revisit the algorithm to get it to work faster with GW52s. On my test machine it takes almost 4 minutes to work out a 1hr result on a large GW52 file.

Well it also takes GPSResults quite a long time to process my GW52 files, but I don't think it's quite 4min.
Thanks for that Dylan, my GW52 Alphas are now good on KA72.

Dylan, I have been having heaps of trouble with KA-72 uploading from my GW not loading tracks plus giving me a low reading on Alpha's compaired to my GT I will email you a file.
Thanks Glynn

Select to expand quote

Glynn Merrit said..
Dylan, I have been having heaps of trouble with KA-72 uploading from my GW not loading tracks plus giving me a low reading on Alpha's compaired to my GT I will email you a file.
Thanks Glynn

PEBBAB ?

[ Problem exists between boom and board ]

Dylan. The issues with processing time will probably only get greater when we start seeing 10hz or more logging rates.

I've updated the site today with significantly faster processing time for all divisions apart from Alphas. (I actually spent a day working on an improved Alpha algorithm, only to find it actually ran slower than before.)

With the GW52 sample file I got from decrepit, processing times for all divisions apart from the Alpha are down to less than 1s apiece (the 1hr alone was taking up to 4 minutes to calculate using my test machine.) Unfortunately, the alpha still takes quite a while (around 30s), but overall the load times for larger files should be much improved.

I have to go back to the drawing board on the Alphas.

great work dylan, your not a glutton for punishment by any chance?

Select to expand quote

decrepit said..
your not a glutton for punishment by any chance?

LOL yes I guess I must be.

Hi Dylan,
Im experiencing difficulties downloading to KA72 (really slow) are you doing work on the site & also getting the info across to GPS TC.
Im using a Canmore.
Cheers Dave

PS the upgrade look great too

All sorted!