Performance Benchmarks

Tracelet’s core Dart algorithms are continuously benchmarked against strict performance thresholds. This ensures that features like Kalman filtering, point-in-polygon geofencing, and battery budgeting remain exceptionally lightweight and do not cause battery drain.

The interactive graph below automatically visualizes the history of key benchmark metrics from recent commits. The data is pulled dynamically from the BENCHMARK.md file tracked in the repository.

Lower values (µs/op) represent better performance.

Loading graph...

Benchmark Descriptions

Benchmark	Description	Hot Path
`kalman_process_single`	Single Kalman filter predict+update cycle	Every GPS fix
`kalman_process_100_fixes`	100 sequential GPS fixes through Kalman filter	Sustained tracking
`kalman_process_1k_fixes`	1000 sequential GPS fixes through Kalman filter	Long session
`kalman_reset`	Reset filter state (Float64List fillRange)	Mode change
`haversine_single`	Single haversine distance calculation	Every GPS fix (3+ calls)
`haversine_1k_pairs`	1000 sequential haversine calculations	Batch processing
`pip_4v`	Point-in-polygon, 4-vertex polygon	Simple geofence
`pip_10v`	Point-in-polygon, 10-vertex polygon	Medium polygon
`pip_50v`	Point-in-polygon, 50-vertex polygon	Complex polygon
`pip_100v`	Point-in-polygon, 100-vertex polygon	Detailed boundary
`pip_500v`	Point-in-polygon, 500-vertex polygon	High-detail polygon
`geofence_eval_10_circular`	Evaluate 10 circular geofences	Small deployment
`geofence_eval_100_circular`	Evaluate 100 circular geofences	Medium deployment
`geofence_eval_500_circular`	Evaluate 500 circular geofences	Large deployment
`geofence_eval_10_polygon_6v`	Evaluate 10 polygon geofences (6 vertices each)	Polygon zones
`geofence_eval_50_polygon_6v`	Evaluate 50 polygon geofences (6 vertices each)	Many polygon zones
`processor_1k_fixes`	Full LocationProcessor pipeline, 1000 fixes	Core tracking loop
`processor_1k_adaptive`	LocationProcessor with adaptive mode, 1000 fixes	Battery-aware tracking
`trip_manager_5k_waypoints`	TripManager accumulating 5000 waypoints	Long trip
`schedule_parse`	Parse a schedule string	Schedule evaluation
`schedule_matches`	Check if time matches a schedule	Schedule evaluation
`schedule_isWithin_5_entries`	Check 5 schedule entries	Multi-schedule
`adaptive_compute`	AdaptiveSamplingEngine single computation	Every GPS fix (adaptive)
`location_fromMap`	Deserialize Location from platform map	Every GPS fix
`location_toMap`	Serialize Location to map	Persistence/HTTP
`location_fromMap_toMap_roundtrip`	Full serialization round-trip	Legacy path
`location_copyWithCoords`	Copy Location with new coords (optimized)	Kalman output
`geofence_fromMap_circular`	Deserialize circular geofence	Geofence loading
`geofence_fromMap_polygon`	Deserialize polygon geofence with vertices	Polygon loading
`delta_encode_10`	Delta-encode batch of 10 locations	HTTP sync
`delta_encode_100`	Delta-encode batch of 100 locations	HTTP sync
`delta_encode_500`	Delta-encode batch of 500 locations	Bulk sync
`delta_decode_10`	Delta-decode batch of 10 locations	Server restore
`delta_decode_100`	Delta-decode batch of 100 locations	Server restore
`delta_decode_500`	Delta-decode batch of 500 locations	Bulk restore
`delta_roundtrip_100`	Full encode→decode round-trip, 100 locations	Correctness path
`battery_budget_single_sample`	Single battery sample processing	Every battery read
`battery_budget_60_samples`	60 samples (1-hour simulation)	Sustained tracking
`battery_budget_heavy_drain`	120 samples with aggressive drain	Worst-case budget
`carbon_trip_100_locations`	Full trip with 100 GPS fixes	Trip completion
`carbon_onLocation`	Per-location carbon accounting	Every GPS fix
`carbon_setActivity`	Activity type switching	Activity change
`carbon_cumulative_report`	Generate cumulative report (10 trips)	Report request
`persist_decider_location`	Location persist decision (all modes)	Every GPS fix
`persist_decider_geofence`	Geofence persist decision (all modes)	Geofence event
`config_fromMap`	Deserialize full Config from map	Config restore
`config_toMap`	Serialize full Config to map	Config persist
`config_roundtrip`	Full Config serialization round-trip	Config update
`state_fromMap`	Deserialize State from map	State restore
`state_toMap`	Serialize State to map	State persist
`route_context_toMap`	Serialize RouteContext to map	Route context attach
`route_context_fromMap`	Deserialize RouteContext from map	Route context restore
`route_context_roundtrip`	Full RouteContext serialization round-trip	Route context update
`sync_body_context_toMap_50`	Serialize SyncBodyContext with 50 locations	Sync body build
`sync_body_context_fromMap_50`	Deserialize SyncBodyContext with 50 locations	Sync body restore
`http_config_ssl_toMap`	Serialize HttpConfig with SSL pinning fields	Config persist
`http_config_ssl_fromMap`	Deserialize HttpConfig with SSL pinning fields	Config restore
`http_config_ssl_roundtrip`	Full HttpConfig+SSL serialization round-trip	Config update

For the full testing methodologies, refer to the BENCHMARK.md file in the Tracelet repository root.