Multi-core support about potreeconverter HOT 7 CLOSED

Closing this here since this issue is a bit outdated, and support for it is coming soon..

from potreeconverter.

Is OpenMP (http://openmp.org/wp/) a feasible option?
Not all compilers support it (notably, LLVM and Borland bcc).

from potreeconverter.

I think it would be more scalable to be able to append points to an existing tree. And then just:
for i in *.las; do ./Potreeconverter --final-tree final_tree $i; done
You could even store the final_tree in a shared folder, and run it in a cluster.

EDIT: I tried with OpenMP and when it doesn't segfault it runs slower than sequentially. Sure it is possible to implement it, but I find the alternatives more efficient.

from potreeconverter.

Beeing able to combine the result of several call to potreeconverter would be nice. Such one can distribute the conversion of a large point cloud given in parts of larger grid.

from potreeconverter.

I also think that parallel conversion of multiple converter instances and merging afterwards could work well.

from potreeconverter.

Are there any advancements in this area? I'm currently working with point clouds in *.laz format over 20 GB each with point clouds up to 60 GB each and it takes ages. I'd also be fine with a workaround and have looked here and there but don't have enough knowledge to even know what to do exactly.

from potreeconverter.

Currently working on a complete rewrite of the converter. It'll be much faster and produce ~3 files instead of thousands to millions of files. Will still take a month or two and with some luck our grant proposal will be accepted, which means we'll get to make it even faster in the future.

from potreeconverter.