[mkgmap-dev] routing & rendering penalty with smaller tiles in splitter

Wed Jun 3 09:14:08 BST 2009

On Wed, Jun 03, 2009 at 02:04:56PM +0800, maning sambale wrote:
> Hi,
> 
> I'm attempting to split the Philippines into smaller tiles using
> splitter.  This is to make rendering faster.
> Currently I've tested with the following:
> 
> --max-nodes=800000
> --max-nodes=400000
> --max-nodes=200000
> 
> 
> On my tests with mapsource and roadtrip, no routing and rendering
> problems so far.  What do others experience when making smaller tiles?
>  What is the "optimum" tile size for very dense data areas with
> minimal penalty for routing and rendering?

I haven't experimented with tile sizes lately.  Months ago, when mkgmap
was in worse shape and I tried routing a 70 km journey through mostly
rural areas (no cycleways), it looked like the Garmin Edge 705 can't look
much across tile borders.  It routed via a major road that ran parallel
to the tile border, and finally turned 90 degrees, causing a 20 km detour.

On a related note, I made a 93 km round-trip bicycle ride last Sunday,
across the Helsinki metropolitan area.  For the first 10+ km, the Edge 705
would tell me to use bicycle roads.  But then it suddenly told me to
switch to the car roads.  I had to recalculate multiple times, and the
route wasn't optimal.

	Marko