-
Notifications
You must be signed in to change notification settings - Fork 1k
Optimization Experiments
As part of our push toward a 1.0 release supporting large regional deployments, we are reducing OTP memory consumption and cold start time. Some optimizations are obviously advantageous, but for others we are not certain whether the improvement in resource consumption will be worth the additional complexity or implementation effort. Therefore, in advance of implementing these optimizations we are carrying out a series of experiments and making minimal prototypes as a sort of ex-ante cost/benefit analysis.
Each of these experiments appears in the OTP Github issue tracker and is tagged as an experiment. During the experiments, results and interpretations are discussed on the OTP-developers mailing list. Once an experiment is completed, the results will be summarized in a wiki page and linked to below.
- Experiment:-Store-all-edges-in-MapDB
- Experiment:-Memory-usage-statistics
- Experiment:-Compact-LineString-geometry
- Experiment:-Hash-Grid-(edge-vertices-stations)-spatial-index
- Experiment:-Azimuth-computation-optimization
- Experiment:-Store-geometry-edges-in-MapDB
- Experiment:-Speed-up-elevation-builder-for-Geotiff (useless for USA)
- Experiment:-Use-fast-serialization-library-for-serializing-graph
- Experiment:-Elevation-data-optimization
- Experiment:-Optimization-of-street-edge-notes
unless you are intentionally working with legacy versions of OpenTripPlanner. Please consult the current documentation at readthedocs