Refactor map_model to store Lanes inside of their parent Roads #747
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Not regenerating yet.
…ds. #746 Somewhat invasive API change internal to map_model, but not much impact elsewhere. Not regenerating anything yet.
Green Lake sim crashes, not investigating. [rebuild] [release]
dabreegster
force-pushed
the
road_embed_lanes
branch
from
215b778 to
2d31d5c
Compare
dabreegster
added a commit
that referenced
this pull request
Oct 8, 2021
by squeezing some start-up performance of huge maps. #746 Shaving at least 1 second off quadtree initialization in huge_seattle from this, no noticeable effects otherwise.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
During #597, we changed from a simple flat
Vec<Lane>to aBTreeMap<LaneID, Lane>, because while editing roads, we might delete or create new lanes. The ID space became non-contiguous. Since then, I've seen some slight perf degradation and lots ofBTreeMaplookups during profiling. #746 and working on adding new routing parameters motivated me to look at this.The idea is similar to how we made
Turnslive inside of their parentIntersectionin #675.LaneIDnow encodes the parent road and the offset from the left side of the road, so looking up a lane becomes two array lookups, not aBTreeMapquery. This change also greatly simplifies a bunch of APIs -- we don't need to pass the wholeMapto compute some stuff about roads, we no longer have this odd duplicatelanes_ltr: Vec<(LaneID, Direction, LaneType)>listing in eachRoad.I'm having trouble with my profiler still, but I benchmarked
make_input_graphfor bike pathfinding. Before this PR, I get around 1.1s, and after, I get around 0.97s. Very slight, but noticeable, improvement. Also a very slight file size savings, since we store less redundant data --huge_seattledrops about 5MB, from 268MB. Definitely more profiling and optimization is needed to address the slowmake_input_graphproblem.Some ideas for future cleanups:
DrawMapstore theDrawLanesnested underDrawRoad, same as the map model. Right now we use aHashMap<LaneID, DrawLane>, which performs fine.lane_center_ptsand lazily calculate it from the road when needed. Probably not worth it; large map files are just a proxy for the real goal of fast startup time on large maps in the web.I'm regenerating everything now, will merge once that's successful.