[geos-devel] Performance of GEOS

[Martin Davis]

Something we may need to brainstorm a bit on is how much attention we should pay to the performance of the code in GEOS.  Currently it seems to perform at about the level of JTS, or a bit slower.  This may well be "good enough" for use in PostGIS - that's something you PostGIS wonks will have to decide.  But there may be room for some real improvements in performance, if we start getting clever about how we interact with the C++ memory manager.  (malloc is notoriously slow, so a bulk allocation scheme might bring some real improvements).

Yury and I are going to do a bit of investigation to make sure we're not overlooking any obvious speedups in the straightforward port.

We need to decide how to prioritize the development - are we looking for functionality first, or good performance from the relate code?

(The fact that "naively implemented" GEOS code is no faster than the Java code is quite interesting - so much for complaints that Java is "too slow" for production work!)

Martin Davis, Senior Technical Specialist
Vivid Solutions Inc.
Suite #1A-2328 Government Street   Victoria, B.C.   V8T 5G5
Phone: (250) 385 6040    Fax: (250) 385 6046
EMail: mbdavis at vividsolutions.com  Web: www.vividsolutions.com