Opened 10 years ago
Closed 10 years ago
#8986 closed enhancement (fixed)
[patch] Data loading/bbox index counting optimization
| Reported by: | shinigami | Owned by: | team |
|---|---|---|---|
| Priority: | major | Milestone: | |
| Component: | Core | Version: | |
| Keywords: | performance | Cc: |
Description
I made few small changes in counting index for bbox and given level, on my testing data and in profiler it seems to run about 6-7x faster now, speedup when loading data is about 20%.
Attachments (1)
Change History (10)
comment:1 Changed 10 years ago by
comment:2 Changed 10 years ago by
| Summary: | Data loading/bbox index counting optimization → [patch] Data loading/bbox index counting optimization |
|---|
comment:3 Changed 10 years ago by
| Keywords: | performance added |
|---|---|
| Priority: | normal → major |
Changed 10 years ago by
| Attachment: | quadbucketsindex.diff added |
|---|
comment:5 Changed 10 years ago by
If I wanted to be nitpicky, I'd say it should be public static int instead of static public int (in regards to modifiers order coding style enhancement) ;)
comment:6 Changed 10 years ago by
How do you measure performance gain ? I'm of course ok to apply this patch but I would be curious to measure it myself before :)
comment:7 Changed 10 years ago by
I run it in profiler and loaded same files (used files from around my city, all ~30000 nodes with some overlaping). Load time felt from 50s to 40s (I have used full method tracking, so it was slow), index counting parts from about 12s to 2s. But it is is biased by profiler overhead..
Nice :) can you also write javadoc for public methods ? Thanks :)