On Sun, Jul 11, 2010 at 1:19 PM, Marcus Wolschon <[email protected]> wrote: > You would you avoid random IO? > On a flash-chip random IO is as cheap as > linear IO and not much slower then RAM > access. (both chips being on the same > memory-bus and no caching or speculative > reading in an ARM CPU)
You can arrange the graph in a way that the data necessary for one route computation is grouped together. This is highly beneficial as you only have to load a few blocks from the external graph file into memory. I am currently working on a mobile ( and Desktop ) routing application using Contraction Hierarchies, which can handle graphs as large as Germany or Europe. As Dennis already announced a beta version will be released during the next weeks as open source. All I can say about the routing performance for now is that is takes less than 50ms and 2MB cache on a typical smart phone to compute a route about 500km long. Greeting, Christian Vetter _______________________________________________ Routing mailing list [email protected] http://lists.openstreetmap.org/listinfo/routing
