search

ngageoint / hootenanny

Hootenanny conflates multiple maps into a single seamless map.
GNU General Public License v3.0
356 stars 74 forks source link

Tune uses of HilbertRTree to speed up accesses to geospatial indexing #3054

Open bwitham opened 5 years ago

bwitham commented 5 years ago

I believe alll the HilbertRTree indexing values have been based on a single trial run against a single dataset. Doing some more intensive testing and against a handful of datasets may end up helping overall runtime performance. Look for refs to this issue in code. For very large datasets, I've noticed for various conflation types where pulling surrounding features from the geospatial index is a performance bottleneck.

bwitham commented 5 years ago

Going to take a shot at this only for the sake of possibly speeding up the conflate run for roads from #3530. If nothing comes out of some short time spent on this, then will shelve it to backlog for the time being since doing this for all tree instances will take a significant amount of time.

bwitham commented 5 years ago

The bottleneck in #3530 doesn't have to do with this indexing, so shelving.

bwitham commented 5 years ago

Actually it might...

bwitham commented 4 years ago

Going to take a look at this, since it is a bottleneck in some conflate scenarios. Not sure if tuning will help or not yet.