Hi Ticker, OK, those numbers look plausible. I see 24 secs without --bounds and 1 minute 49 seconds with --bounds. The latter value goes down to 1 minute 19 seconds when I use -Xmx2G I've tried my idea reg. splitting of the complex boundary relations and found no effect on runtime, only a small increase in the bounds.zip. Gerd ________________________________________ Von: mkgmap-dev <mkgmap-dev-bounces@lists.mkgmap.org.uk> im Auftrag von Ticker Berkin <rwb-mkgmap@jagit.co.uk> Gesendet: Donnerstag, 19. März 2020 15:47 An: Development list for mkgmap Betreff: Re: [mkgmap-dev] Performance with --bounds option Hi Gerd These are the timings I get for the single tile 74220001: trunk with --bounds=bounds.zip Total time taken: 3 minutes 44 seconds r4295 with --bounds=bounds.zip Total time taken: 3 minutes 51 seconds trunk with --no-bounds Total time taken: 1 minute 11 seconds r4295 with --no-bounds Total time taken: 1 minute 6 seconds so the latest version is about the same speed as versions before the is_in merge. What I meant by 'acceptable' was that the extra time for bounds processing, although bad for this tile, is not a concern. Building britain-and-ireland with bounds takes 36 minutes, without it takes 29 minutes. Ticker On Thu, 2020-03-19 at 12:46 +0000, Gerd Petermann wrote:
Hi Ticker,
when you say acceptable does that mean that latest mkgmap is still much slower than e.g. r4290?
Gerd
________________________________________ Von: mkgmap-dev <mkgmap-dev-bounces@lists.mkgmap.org.uk> im Auftrag von Ticker Berkin <rwb-mkgmap@jagit.co.uk> Gesendet: Donnerstag, 19. März 2020 11:04 An: Development list for mkgmap Betreff: Re: [mkgmap-dev] Performance with --bounds option
Hi Gerd
With the reversion of the code in r4469, the performance is acceptable again. It is only a few tiles that have this problem and, for GB, they don't add significantly to the overall map generation time
Ticker
On Thu, 2020-03-19 at 06:52 +0000, Gerd Petermann wrote:
Hi Ticker,
some more details: I didn't think of boundary relations like r1959008 which contains > 700 outer rings. For each node close to that boundary mkgmap called Java2DConverter.areaToSingularAreas() which calculated all the areas again and again. In Germany, boundary relations typically have only one ring ;) I think I should look into the code which compiles bounds.zip, somehow I expected that those complex areas are split to singular areas, might improve performance.
Gerd
________________________________________ Von: mkgmap-dev <mkgmap-dev-bounces@lists.mkgmap.org.uk> im Auftrag von Gerd Petermann <gpetermann_muenchen@hotmail.com> Gesendet: Mittwoch, 18. März 2020 19:46 An: Development list for mkgmap Betreff: Re: [mkgmap-dev] Performance with --bounds option
Hi Ticker,
I forgot to remove some experimental code for is-in branch in BoundaryQuadTree. Fixed with r4469. I'll probably also remove the methods pointInsideArea() and pointInsideSingularArea() in BoundaryUtil.
Gerd
________________________________________ Von: mkgmap-dev <mkgmap-dev-bounces@lists.mkgmap.org.uk> im Auftrag von Ticker Berkin <rwb-mkgmap@jagit.co.uk> Gesendet: Mittwoch, 18. März 2020 18:16 An: Development list for mkgmap Betreff: Re: [mkgmap-dev] Performance with --bounds option
Have done:
http://files.mkgmap.org.uk/download/462/74220001.osm.pbf
Ticker
On Wed, 2020-03-18 at 16:26 +0000, Gerd Petermann wrote:
Hi Ticker,
will look at it tomorrow. Maybe you can upload the file to http://files.mkgmap.org.uk/
Gerd
________________________________________ Von: mkgmap-dev <mkgmap-dev-bounces@lists.mkgmap.org.uk> im Auftrag von Ticker Berkin <rwb-mkgmap@jagit.co.uk> Gesendet: Mittwoch, 18. März 2020 11:56 An: Development list for mkgmap Betreff: Re: [mkgmap-dev] Performance with --bounds option
Hi Gerd
It's
74220001: 2674688,-700416 to 2811904,-233472 # : 57.392578,-15.029297 to 60.336914,-5.009766
Ticker
On Wed, 2020-03-18 at 10:46 +0000, Gerd Petermann wrote:
Hi Ticker,
please tell me the tile boundaries
Gerd
________________________________________ Von: mkgmap-dev <mkgmap-dev-bounces@lists.mkgmap.org.uk> im Auftrag von Ticker Berkin <rwb-mkgmap@jagit.co.uk> Gesendet: Mittwoch, 18. März 2020 11:43 An: mkgmap development Betreff: [mkgmap-dev] Performance with --bounds option
Hi Gerd
With the current version I have a tile that takes 3.5 hours to build. With an old version (r4295) it took about 3 minutes. Without - -bounds it takes 1 minute 15 secs.
Relevant options are: --bounds=bounds.zip --location-autofill=is_in,nearest
From the mkgmap.log files: INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: Checking bounds dir took 66 ms ... INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: Starting with location hook INFO: uk.me.parabola.mkgmap.reader.osm.boundary.BoundaryUtil ../mapGB/74220001.osm.pbf: loading boundary file: bounds_2650000_ -750000.bnd ... about 40 similar messages ... INFO: uk.me.parabola.mkgmap.reader.osm.boundary.BoundaryUtil ../mapGB/74220001.osm.pbf: loading boundary file: bounds_2800000_ -250000.bnd WARN: uk.me.parabola.mkgmap.reader.osm.boundary.BoundaryGrid ../mapGB/74220001.osm.pbf: no precompiled boundary information available for raster tile 2750000_-400000 ... 16 similar messages ... WARN: uk.me.parabola.mkgmap.reader.osm.boundary.BoundaryGrid ../mapGB/74220001.osm.pbf: no precompiled boundary information available for raster tile 2800000_-250000 INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: ======= LocationHook Stats ===== INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: QuadTree searches : 97834 INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: unsuccesfull : 20 INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: unsuccesfull for ways: 9 INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: Location hook finished in 12747318 ms
The old version had similar lines, but the stats at the end are:
INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: ======= LocationHook Stats ===== INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: QuadTree searches : 95639 INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: unsuccesfull : 20 INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: unsuccesfull for ways: 9 INFO: uk.me.parabola.mkgmap.reader.osm.LocationHook ../mapGB/74220001.osm.pbf: Location hook finished in 113704 ms
Ticker
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev